llm.rb 11.3.1 → 12.1.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (76) hide show
  1. checksums.yaml +4 -4
  2. data/CHANGELOG.md +354 -2
  3. data/LICENSE +93 -17
  4. data/README.md +226 -616
  5. data/data/anthropic.json +322 -426
  6. data/data/bedrock.json +2634 -1144
  7. data/data/deepinfra.json +1513 -0
  8. data/data/deepseek.json +57 -28
  9. data/data/google.json +411 -771
  10. data/data/openai.json +1104 -771
  11. data/data/xai.json +141 -292
  12. data/data/zai.json +263 -141
  13. data/lib/llm/active_record/acts_as_agent.rb +3 -41
  14. data/lib/llm/active_record/acts_as_llm.rb +18 -0
  15. data/lib/llm/active_record.rb +3 -3
  16. data/lib/llm/agent.rb +25 -1
  17. data/lib/llm/context.rb +17 -5
  18. data/lib/llm/contract/completion.rb +2 -2
  19. data/lib/llm/json_adapter.rb +29 -3
  20. data/lib/llm/provider.rb +3 -3
  21. data/lib/llm/providers/deepinfra/audio.rb +66 -0
  22. data/lib/llm/providers/deepinfra/images.rb +90 -0
  23. data/lib/llm/providers/deepinfra/response_adapter.rb +36 -0
  24. data/lib/llm/providers/deepinfra.rb +100 -0
  25. data/lib/llm/providers/deepseek/images.rb +109 -0
  26. data/lib/llm/providers/deepseek/request_adapter.rb +32 -0
  27. data/lib/llm/providers/deepseek/response_adapter/image.rb +9 -0
  28. data/lib/llm/providers/deepseek/response_adapter.rb +29 -0
  29. data/lib/llm/providers/deepseek.rb +4 -2
  30. data/lib/llm/providers/google/request_adapter.rb +22 -5
  31. data/lib/llm/providers/google.rb +4 -4
  32. data/lib/llm/providers/llamacpp.rb +5 -5
  33. data/lib/llm/providers/openai/audio.rb +6 -2
  34. data/lib/llm/providers/openai/images.rb +9 -50
  35. data/lib/llm/providers/openai/request_adapter/respond.rb +38 -4
  36. data/lib/llm/providers/openai/response_adapter/audio.rb +5 -1
  37. data/lib/llm/providers/openai/response_adapter/completion.rb +1 -1
  38. data/lib/llm/providers/openai/response_adapter/image.rb +0 -4
  39. data/lib/llm/providers/openai/responses.rb +1 -0
  40. data/lib/llm/providers/openai/stream_parser.rb +5 -6
  41. data/lib/llm/providers/openai.rb +2 -2
  42. data/lib/llm/providers/xai/images.rb +49 -26
  43. data/lib/llm/providers/xai.rb +2 -2
  44. data/lib/llm/repl/input.rb +64 -0
  45. data/lib/llm/repl/status.rb +30 -0
  46. data/lib/llm/repl/stream.rb +46 -0
  47. data/lib/llm/repl/transcript.rb +61 -0
  48. data/lib/llm/repl/window.rb +107 -0
  49. data/lib/llm/repl.rb +78 -0
  50. data/lib/llm/response.rb +10 -0
  51. data/lib/llm/schema/leaf.rb +7 -1
  52. data/lib/llm/schema/renderer.rb +121 -0
  53. data/lib/llm/schema.rb +30 -0
  54. data/lib/llm/sequel/agent.rb +2 -43
  55. data/lib/llm/sequel/plugin.rb +25 -7
  56. data/lib/llm/tools/chdir.rb +23 -0
  57. data/lib/llm/tools/git.rb +41 -0
  58. data/lib/llm/tools/mkdir.rb +32 -0
  59. data/lib/llm/tools/pwd.rb +20 -0
  60. data/lib/llm/tools/read_file.rb +40 -0
  61. data/lib/llm/tools/rg.rb +46 -0
  62. data/lib/llm/tools/shell.rb +48 -0
  63. data/lib/llm/tools/swap_text.rb +25 -0
  64. data/lib/llm/tools/write_file.rb +24 -0
  65. data/lib/llm/tools.rb +5 -0
  66. data/lib/llm/tracer/telemetry.rb +4 -6
  67. data/lib/llm/tracer.rb +9 -21
  68. data/lib/llm/transport/execution.rb +16 -1
  69. data/lib/llm/transport/net_http_adapter.rb +1 -1
  70. data/lib/llm/uridata.rb +16 -0
  71. data/lib/llm/version.rb +1 -1
  72. data/lib/llm.rb +9 -0
  73. data/llm.gemspec +5 -18
  74. data/resources/deepdive.md +829 -263
  75. metadata +31 -18
  76. data/lib/llm/tracer/langsmith.rb +0 -144
data/data/bedrock.json CHANGED
@@ -10,45 +10,60 @@
10
10
  "name": "Amazon Bedrock",
11
11
  "doc": "https://docs.aws.amazon.com/bedrock/latest/userguide/models-supported.html",
12
12
  "models": {
13
- "zai.glm-4.7": {
14
- "id": "zai.glm-4.7",
15
- "name": "GLM-4.7",
16
- "family": "glm",
17
- "attachment": false,
13
+ "global.anthropic.claude-haiku-4-5-20251001-v1:0": {
14
+ "id": "global.anthropic.claude-haiku-4-5-20251001-v1:0",
15
+ "name": "Claude Haiku 4.5 (Global)",
16
+ "description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
17
+ "family": "claude-haiku",
18
+ "attachment": true,
18
19
  "reasoning": true,
20
+ "reasoning_options": [
21
+ {
22
+ "type": "budget_tokens",
23
+ "min": 1024
24
+ }
25
+ ],
19
26
  "tool_call": true,
20
- "interleaved": {
21
- "field": "reasoning_content"
22
- },
23
27
  "structured_output": true,
24
28
  "temperature": true,
25
- "knowledge": "2025-04",
26
- "release_date": "2025-12-22",
27
- "last_updated": "2025-12-22",
29
+ "knowledge": "2025-02-28",
30
+ "release_date": "2025-10-15",
31
+ "last_updated": "2025-10-15",
28
32
  "modalities": {
29
33
  "input": [
30
- "text"
34
+ "text",
35
+ "image",
36
+ "pdf"
31
37
  ],
32
38
  "output": [
33
39
  "text"
34
40
  ]
35
41
  },
36
- "open_weights": true,
42
+ "open_weights": false,
37
43
  "limit": {
38
- "context": 204800,
39
- "output": 131072
44
+ "context": 200000,
45
+ "output": 64000
40
46
  },
41
47
  "cost": {
42
- "input": 0.6,
43
- "output": 2.2
48
+ "input": 1,
49
+ "output": 5,
50
+ "cache_read": 0.1,
51
+ "cache_write": 1.25
44
52
  }
45
53
  },
46
54
  "global.anthropic.claude-sonnet-4-5-20250929-v1:0": {
47
55
  "id": "global.anthropic.claude-sonnet-4-5-20250929-v1:0",
48
56
  "name": "Claude Sonnet 4.5 (Global)",
57
+ "description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
49
58
  "family": "claude-sonnet",
50
59
  "attachment": true,
51
60
  "reasoning": true,
61
+ "reasoning_options": [
62
+ {
63
+ "type": "budget_tokens",
64
+ "min": 1024
65
+ }
66
+ ],
52
67
  "tool_call": true,
53
68
  "structured_output": true,
54
69
  "temperature": true,
@@ -77,21 +92,22 @@
77
92
  "cache_write": 3.75
78
93
  }
79
94
  },
80
- "qwen.qwen3-235b-a22b-2507-v1:0": {
81
- "id": "qwen.qwen3-235b-a22b-2507-v1:0",
82
- "name": "Qwen3 235B A22B 2507",
83
- "family": "qwen",
84
- "attachment": false,
95
+ "us.meta.llama4-scout-17b-instruct-v1:0": {
96
+ "id": "us.meta.llama4-scout-17b-instruct-v1:0",
97
+ "name": "Llama 4 Scout 17B Instruct (US)",
98
+ "description": "Open Llama with long-context vision for efficient multimodal agents",
99
+ "family": "llama",
100
+ "attachment": true,
85
101
  "reasoning": false,
86
102
  "tool_call": true,
87
- "structured_output": true,
88
103
  "temperature": true,
89
- "knowledge": "2024-04",
90
- "release_date": "2025-09-18",
91
- "last_updated": "2025-09-18",
104
+ "knowledge": "2024-08",
105
+ "release_date": "2025-04-05",
106
+ "last_updated": "2025-04-05",
92
107
  "modalities": {
93
108
  "input": [
94
- "text"
109
+ "text",
110
+ "image"
95
111
  ],
96
112
  "output": [
97
113
  "text"
@@ -99,28 +115,27 @@
99
115
  },
100
116
  "open_weights": true,
101
117
  "limit": {
102
- "context": 262144,
103
- "output": 131072
118
+ "context": 3500000,
119
+ "output": 16384
104
120
  },
105
121
  "cost": {
106
- "input": 0.22,
107
- "output": 0.88
122
+ "input": 0.17,
123
+ "output": 0.66
108
124
  }
109
125
  },
110
- "zai.glm-5": {
111
- "id": "zai.glm-5",
112
- "name": "GLM-5",
113
- "family": "glm",
126
+ "minimax.minimax-m2": {
127
+ "id": "minimax.minimax-m2",
128
+ "name": "MiniMax M2",
129
+ "description": "MiniMax model for chat, coding, office work, and agentic tasks",
130
+ "family": "minimax",
114
131
  "attachment": false,
115
132
  "reasoning": true,
133
+ "reasoning_options": [],
116
134
  "tool_call": true,
117
- "interleaved": {
118
- "field": "reasoning_content"
119
- },
120
- "structured_output": true,
135
+ "structured_output": false,
121
136
  "temperature": true,
122
- "release_date": "2026-03-18",
123
- "last_updated": "2026-03-18",
137
+ "release_date": "2025-10-27",
138
+ "last_updated": "2025-10-27",
124
139
  "modalities": {
125
140
  "input": [
126
141
  "text"
@@ -131,60 +146,43 @@
131
146
  },
132
147
  "open_weights": true,
133
148
  "limit": {
134
- "context": 202752,
135
- "output": 101376
149
+ "context": 204608,
150
+ "output": 128000
136
151
  },
137
152
  "cost": {
138
- "input": 1,
139
- "output": 3.2
153
+ "input": 0.3,
154
+ "output": 1.2
140
155
  }
141
156
  },
142
- "amazon.nova-pro-v1:0": {
143
- "id": "amazon.nova-pro-v1:0",
144
- "name": "Nova Pro",
145
- "family": "nova-pro",
157
+ "anthropic.claude-opus-4-7": {
158
+ "id": "anthropic.claude-opus-4-7",
159
+ "name": "Claude Opus 4.7",
160
+ "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
161
+ "family": "claude-opus",
146
162
  "attachment": true,
147
- "reasoning": false,
163
+ "reasoning": true,
164
+ "reasoning_options": [
165
+ {
166
+ "type": "effort",
167
+ "values": [
168
+ "low",
169
+ "medium",
170
+ "high",
171
+ "xhigh",
172
+ "max"
173
+ ]
174
+ }
175
+ ],
148
176
  "tool_call": true,
149
- "temperature": true,
150
- "knowledge": "2024-10",
151
- "release_date": "2024-12-03",
152
- "last_updated": "2024-12-03",
177
+ "temperature": false,
178
+ "knowledge": "2026-01-31",
179
+ "release_date": "2026-04-16",
180
+ "last_updated": "2026-04-16",
153
181
  "modalities": {
154
182
  "input": [
155
183
  "text",
156
184
  "image",
157
- "video"
158
- ],
159
- "output": [
160
- "text"
161
- ]
162
- },
163
- "open_weights": false,
164
- "limit": {
165
- "context": 300000,
166
- "output": 8192
167
- },
168
- "cost": {
169
- "input": 0.8,
170
- "output": 3.2,
171
- "cache_read": 0.2
172
- }
173
- },
174
- "openai.gpt-oss-20b-1:0": {
175
- "id": "openai.gpt-oss-20b-1:0",
176
- "name": "gpt-oss-20b",
177
- "family": "gpt-oss",
178
- "attachment": false,
179
- "reasoning": false,
180
- "tool_call": true,
181
- "structured_output": true,
182
- "temperature": true,
183
- "release_date": "2025-08-05",
184
- "last_updated": "2025-08-05",
185
- "modalities": {
186
- "input": [
187
- "text"
185
+ "pdf"
188
186
  ],
189
187
  "output": [
190
188
  "text"
@@ -192,26 +190,44 @@
192
190
  },
193
191
  "open_weights": false,
194
192
  "limit": {
195
- "context": 128000,
196
- "output": 16384
193
+ "context": 1000000,
194
+ "output": 128000
197
195
  },
198
196
  "cost": {
199
- "input": 0.07,
200
- "output": 0.3
197
+ "input": 5,
198
+ "output": 25,
199
+ "cache_read": 0.5,
200
+ "cache_write": 6.25
201
201
  }
202
202
  },
203
- "us.anthropic.claude-haiku-4-5-20251001-v1:0": {
204
- "id": "us.anthropic.claude-haiku-4-5-20251001-v1:0",
205
- "name": "Claude Haiku 4.5 (US)",
206
- "family": "claude-haiku",
203
+ "eu.anthropic.claude-sonnet-4-6": {
204
+ "id": "eu.anthropic.claude-sonnet-4-6",
205
+ "name": "Claude Sonnet 4.6 (EU)",
206
+ "description": "Claude workhorse for coding agents, careful analysis, and production cost control",
207
+ "family": "claude-sonnet",
207
208
  "attachment": true,
208
209
  "reasoning": true,
210
+ "reasoning_options": [
211
+ {
212
+ "type": "effort",
213
+ "values": [
214
+ "low",
215
+ "medium",
216
+ "high",
217
+ "max"
218
+ ]
219
+ },
220
+ {
221
+ "type": "budget_tokens",
222
+ "min": 1024
223
+ }
224
+ ],
209
225
  "tool_call": true,
210
226
  "structured_output": true,
211
227
  "temperature": true,
212
- "knowledge": "2025-02-28",
213
- "release_date": "2025-10-15",
214
- "last_updated": "2025-10-15",
228
+ "knowledge": "2025-08-31",
229
+ "release_date": "2026-02-17",
230
+ "last_updated": "2026-03-13",
215
231
  "modalities": {
216
232
  "input": [
217
233
  "text",
@@ -224,59 +240,63 @@
224
240
  },
225
241
  "open_weights": false,
226
242
  "limit": {
227
- "context": 200000,
243
+ "context": 1000000,
228
244
  "output": 64000
229
245
  },
230
246
  "cost": {
231
- "input": 1,
232
- "output": 5,
233
- "cache_read": 0.1,
234
- "cache_write": 1.25
247
+ "input": 3.3,
248
+ "output": 16.5,
249
+ "cache_read": 0.33,
250
+ "cache_write": 4.125
235
251
  }
236
252
  },
237
- "writer.palmyra-x4-v1:0": {
238
- "id": "writer.palmyra-x4-v1:0",
239
- "name": "Palmyra X4",
240
- "family": "palmyra",
241
- "attachment": false,
242
- "reasoning": true,
253
+ "mistral.voxtral-small-24b-2507": {
254
+ "id": "mistral.voxtral-small-24b-2507",
255
+ "name": "Voxtral Small 24B 2507",
256
+ "description": "Efficient Mistral model for fast chat, extraction, and production assistants",
257
+ "family": "mistral",
258
+ "attachment": true,
259
+ "reasoning": false,
243
260
  "tool_call": true,
261
+ "structured_output": true,
244
262
  "temperature": true,
245
- "release_date": "2025-04-28",
246
- "last_updated": "2025-04-28",
263
+ "release_date": "2025-07-01",
264
+ "last_updated": "2025-07-01",
247
265
  "modalities": {
248
266
  "input": [
249
- "text"
267
+ "text",
268
+ "audio"
250
269
  ],
251
270
  "output": [
252
271
  "text"
253
272
  ]
254
273
  },
255
- "open_weights": false,
274
+ "open_weights": true,
256
275
  "limit": {
257
- "context": 122880,
276
+ "context": 32000,
258
277
  "output": 8192
259
278
  },
260
279
  "cost": {
261
- "input": 2.5,
262
- "output": 10
280
+ "input": 0.15,
281
+ "output": 0.35
263
282
  }
264
283
  },
265
- "moonshot.kimi-k2-thinking": {
266
- "id": "moonshot.kimi-k2-thinking",
267
- "name": "Kimi K2 Thinking",
268
- "family": "kimi-thinking",
284
+ "mistral.ministral-3-3b-instruct": {
285
+ "id": "mistral.ministral-3-3b-instruct",
286
+ "name": "Ministral 3 3B",
287
+ "description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
288
+ "family": "ministral",
269
289
  "attachment": false,
270
- "reasoning": true,
290
+ "reasoning": false,
271
291
  "tool_call": true,
272
- "interleaved": true,
273
292
  "structured_output": true,
274
293
  "temperature": true,
275
294
  "release_date": "2025-12-02",
276
295
  "last_updated": "2025-12-02",
277
296
  "modalities": {
278
297
  "input": [
279
- "text"
298
+ "text",
299
+ "image"
280
300
  ],
281
301
  "output": [
282
302
  "text"
@@ -285,24 +305,35 @@
285
305
  "open_weights": true,
286
306
  "limit": {
287
307
  "context": 256000,
288
- "output": 256000
308
+ "output": 8192
289
309
  },
290
310
  "cost": {
291
- "input": 0.6,
292
- "output": 2.5
311
+ "input": 0.1,
312
+ "output": 0.1
293
313
  }
294
314
  },
295
- "mistral.ministral-3-8b-instruct": {
296
- "id": "mistral.ministral-3-8b-instruct",
297
- "name": "Ministral 3 8B",
298
- "family": "ministral",
315
+ "openai.gpt-oss-20b": {
316
+ "id": "openai.gpt-oss-20b",
317
+ "name": "gpt-oss-20b",
318
+ "description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
319
+ "family": "gpt-oss",
299
320
  "attachment": false,
300
- "reasoning": false,
321
+ "reasoning": true,
322
+ "reasoning_options": [
323
+ {
324
+ "type": "effort",
325
+ "values": [
326
+ "low",
327
+ "medium",
328
+ "high"
329
+ ]
330
+ }
331
+ ],
301
332
  "tool_call": true,
302
333
  "structured_output": true,
303
334
  "temperature": true,
304
- "release_date": "2024-12-01",
305
- "last_updated": "2024-12-01",
335
+ "release_date": "2025-08-05",
336
+ "last_updated": "2025-08-05",
306
337
  "modalities": {
307
338
  "input": [
308
339
  "text"
@@ -314,49 +345,119 @@
314
345
  "open_weights": false,
315
346
  "limit": {
316
347
  "context": 128000,
317
- "output": 4096
348
+ "output": 16384
349
+ },
350
+ "provider": {
351
+ "npm": "@ai-sdk/amazon-bedrock/mantle",
352
+ "api": "https://bedrock-mantle.${AWS_REGION}.api.aws/v1",
353
+ "shape": "responses"
318
354
  },
319
355
  "cost": {
320
- "input": 0.15,
321
- "output": 0.15
356
+ "input": 0.07,
357
+ "output": 0.3
322
358
  }
323
359
  },
324
- "meta.llama4-scout-17b-instruct-v1:0": {
325
- "id": "meta.llama4-scout-17b-instruct-v1:0",
326
- "name": "Llama 4 Scout 17B Instruct",
327
- "family": "llama",
360
+ "anthropic.claude-opus-4-6-v1": {
361
+ "id": "anthropic.claude-opus-4-6-v1",
362
+ "name": "Claude Opus 4.6",
363
+ "description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
364
+ "family": "claude-opus",
328
365
  "attachment": true,
329
- "reasoning": false,
366
+ "reasoning": true,
367
+ "reasoning_options": [
368
+ {
369
+ "type": "effort",
370
+ "values": [
371
+ "low",
372
+ "medium",
373
+ "high",
374
+ "max"
375
+ ]
376
+ },
377
+ {
378
+ "type": "budget_tokens",
379
+ "min": 1024
380
+ }
381
+ ],
330
382
  "tool_call": true,
383
+ "structured_output": true,
331
384
  "temperature": true,
332
- "knowledge": "2024-08",
333
- "release_date": "2025-04-05",
334
- "last_updated": "2025-04-05",
385
+ "knowledge": "2025-05-31",
386
+ "release_date": "2026-02-05",
387
+ "last_updated": "2026-03-13",
335
388
  "modalities": {
336
389
  "input": [
337
390
  "text",
338
- "image"
391
+ "image",
392
+ "pdf"
339
393
  ],
340
394
  "output": [
341
395
  "text"
342
396
  ]
343
397
  },
344
- "open_weights": true,
398
+ "open_weights": false,
345
399
  "limit": {
346
- "context": 3500000,
400
+ "context": 1000000,
401
+ "output": 128000
402
+ },
403
+ "cost": {
404
+ "input": 5,
405
+ "output": 25,
406
+ "cache_read": 0.5,
407
+ "cache_write": 6.25
408
+ }
409
+ },
410
+ "openai.gpt-oss-safeguard-20b": {
411
+ "id": "openai.gpt-oss-safeguard-20b",
412
+ "name": "GPT OSS Safeguard 20B",
413
+ "description": "Safety model for policy screening, moderation, and risk-aware routing workflows",
414
+ "family": "gpt-oss",
415
+ "attachment": false,
416
+ "reasoning": false,
417
+ "tool_call": true,
418
+ "structured_output": true,
419
+ "temperature": true,
420
+ "release_date": "2025-10-29",
421
+ "last_updated": "2025-10-29",
422
+ "modalities": {
423
+ "input": [
424
+ "text"
425
+ ],
426
+ "output": [
427
+ "text"
428
+ ]
429
+ },
430
+ "open_weights": false,
431
+ "limit": {
432
+ "context": 128000,
347
433
  "output": 16384
348
434
  },
349
435
  "cost": {
350
- "input": 0.17,
351
- "output": 0.66
436
+ "input": 0.07,
437
+ "output": 0.2
352
438
  }
353
439
  },
354
- "us.anthropic.claude-opus-4-5-20251101-v1:0": {
355
- "id": "us.anthropic.claude-opus-4-5-20251101-v1:0",
356
- "name": "Claude Opus 4.5 (US)",
440
+ "anthropic.claude-opus-4-5-20251101-v1:0": {
441
+ "id": "anthropic.claude-opus-4-5-20251101-v1:0",
442
+ "name": "Claude Opus 4.5",
443
+ "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
357
444
  "family": "claude-opus",
358
445
  "attachment": true,
359
446
  "reasoning": true,
447
+ "reasoning_options": [
448
+ {
449
+ "type": "effort",
450
+ "values": [
451
+ "low",
452
+ "medium",
453
+ "high"
454
+ ]
455
+ },
456
+ {
457
+ "type": "budget_tokens",
458
+ "min": 1024
459
+ }
460
+ ],
360
461
  "tool_call": true,
361
462
  "structured_output": true,
362
463
  "temperature": true,
@@ -385,78 +486,74 @@
385
486
  "cache_write": 6.25
386
487
  }
387
488
  },
388
- "qwen.qwen3-coder-480b-a35b-v1:0": {
389
- "id": "qwen.qwen3-coder-480b-a35b-v1:0",
390
- "name": "Qwen3 Coder 480B A35B Instruct",
391
- "family": "qwen",
392
- "attachment": false,
393
- "reasoning": false,
394
- "tool_call": true,
395
- "structured_output": true,
396
- "temperature": true,
397
- "knowledge": "2024-04",
398
- "release_date": "2025-09-18",
399
- "last_updated": "2025-09-18",
400
- "modalities": {
401
- "input": [
402
- "text"
403
- ],
404
- "output": [
405
- "text"
406
- ]
407
- },
408
- "open_weights": true,
409
- "limit": {
410
- "context": 131072,
411
- "output": 65536
412
- },
413
- "cost": {
414
- "input": 0.22,
415
- "output": 1.8
416
- }
417
- },
418
- "google.gemma-3-27b-it": {
419
- "id": "google.gemma-3-27b-it",
420
- "name": "Google Gemma 3 27B Instruct",
421
- "family": "gemma",
489
+ "global.anthropic.claude-fable-5": {
490
+ "id": "global.anthropic.claude-fable-5",
491
+ "name": "Claude Fable 5 (Global)",
492
+ "description": "Claude model for creative writing, analysis, and controlled agent workflows",
493
+ "family": "claude-fable",
422
494
  "attachment": true,
423
- "reasoning": false,
495
+ "reasoning": true,
496
+ "reasoning_options": [
497
+ {
498
+ "type": "effort",
499
+ "values": [
500
+ "low",
501
+ "medium",
502
+ "high",
503
+ "xhigh",
504
+ "max"
505
+ ]
506
+ }
507
+ ],
424
508
  "tool_call": true,
425
- "structured_output": true,
426
- "temperature": true,
427
- "knowledge": "2025-07",
428
- "release_date": "2025-07-27",
429
- "last_updated": "2025-07-27",
509
+ "temperature": false,
510
+ "knowledge": "2026-01-31",
511
+ "release_date": "2026-06-09",
512
+ "last_updated": "2026-06-09",
430
513
  "modalities": {
431
514
  "input": [
432
515
  "text",
433
- "image"
516
+ "image",
517
+ "pdf"
434
518
  ],
435
519
  "output": [
436
520
  "text"
437
521
  ]
438
522
  },
439
- "open_weights": true,
523
+ "open_weights": false,
440
524
  "limit": {
441
- "context": 202752,
442
- "output": 8192
525
+ "context": 1000000,
526
+ "output": 128000
443
527
  },
444
528
  "cost": {
445
- "input": 0.12,
446
- "output": 0.2
529
+ "input": 10,
530
+ "output": 50,
531
+ "cache_read": 1,
532
+ "cache_write": 12.5
447
533
  }
448
534
  },
449
- "mistral.ministral-3-14b-instruct": {
450
- "id": "mistral.ministral-3-14b-instruct",
451
- "name": "Ministral 14B 3.0",
452
- "family": "ministral",
535
+ "openai.gpt-oss-120b-1:0": {
536
+ "id": "openai.gpt-oss-120b-1:0",
537
+ "name": "gpt-oss-120b",
538
+ "description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
539
+ "family": "gpt-oss",
453
540
  "attachment": false,
454
- "reasoning": false,
541
+ "reasoning": true,
542
+ "reasoning_options": [
543
+ {
544
+ "type": "effort",
545
+ "values": [
546
+ "low",
547
+ "medium",
548
+ "high"
549
+ ]
550
+ }
551
+ ],
455
552
  "tool_call": true,
456
553
  "structured_output": true,
457
554
  "temperature": true,
458
- "release_date": "2024-12-01",
459
- "last_updated": "2024-12-01",
555
+ "release_date": "2025-08-05",
556
+ "last_updated": "2025-08-05",
460
557
  "modalities": {
461
558
  "input": [
462
559
  "text"
@@ -468,19 +565,26 @@
468
565
  "open_weights": false,
469
566
  "limit": {
470
567
  "context": 128000,
471
- "output": 4096
568
+ "output": 16384
472
569
  },
473
570
  "cost": {
474
- "input": 0.2,
475
- "output": 0.2
571
+ "input": 0.15,
572
+ "output": 0.6
476
573
  }
477
574
  },
478
- "us.anthropic.claude-sonnet-4-5-20250929-v1:0": {
479
- "id": "us.anthropic.claude-sonnet-4-5-20250929-v1:0",
480
- "name": "Claude Sonnet 4.5 (US)",
575
+ "anthropic.claude-sonnet-4-5-20250929-v1:0": {
576
+ "id": "anthropic.claude-sonnet-4-5-20250929-v1:0",
577
+ "name": "Claude Sonnet 4.5",
578
+ "description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
481
579
  "family": "claude-sonnet",
482
580
  "attachment": true,
483
581
  "reasoning": true,
582
+ "reasoning_options": [
583
+ {
584
+ "type": "budget_tokens",
585
+ "min": 1024
586
+ }
587
+ ],
484
588
  "tool_call": true,
485
589
  "structured_output": true,
486
590
  "temperature": true,
@@ -509,18 +613,52 @@
509
613
  "cache_write": 3.75
510
614
  }
511
615
  },
512
- "qwen.qwen3-coder-30b-a3b-v1:0": {
513
- "id": "qwen.qwen3-coder-30b-a3b-v1:0",
514
- "name": "Qwen3 Coder 30B A3B Instruct",
616
+ "amazon.nova-pro-v1:0": {
617
+ "id": "amazon.nova-pro-v1:0",
618
+ "name": "Nova Pro",
619
+ "description": "Flagship model for demanding analysis, coding, and production agent workflows",
620
+ "family": "nova-pro",
621
+ "attachment": true,
622
+ "reasoning": false,
623
+ "tool_call": true,
624
+ "temperature": true,
625
+ "knowledge": "2024-10",
626
+ "release_date": "2024-12-03",
627
+ "last_updated": "2024-12-03",
628
+ "modalities": {
629
+ "input": [
630
+ "text",
631
+ "image",
632
+ "video"
633
+ ],
634
+ "output": [
635
+ "text"
636
+ ]
637
+ },
638
+ "open_weights": false,
639
+ "limit": {
640
+ "context": 300000,
641
+ "output": 8192
642
+ },
643
+ "cost": {
644
+ "input": 0.8,
645
+ "output": 3.2,
646
+ "cache_read": 0.2
647
+ }
648
+ },
649
+ "qwen.qwen3-coder-next": {
650
+ "id": "qwen.qwen3-coder-next",
651
+ "name": "Qwen3 Coder Next",
652
+ "description": "Qwen coding model for software agents, repository edits, and code reasoning",
515
653
  "family": "qwen",
516
654
  "attachment": false,
517
- "reasoning": false,
655
+ "reasoning": true,
656
+ "reasoning_options": [],
518
657
  "tool_call": true,
519
658
  "structured_output": true,
520
659
  "temperature": true,
521
- "knowledge": "2024-04",
522
- "release_date": "2025-09-18",
523
- "last_updated": "2025-09-18",
660
+ "release_date": "2026-02-06",
661
+ "last_updated": "2026-02-06",
524
662
  "modalities": {
525
663
  "input": [
526
664
  "text"
@@ -529,22 +667,35 @@
529
667
  "text"
530
668
  ]
531
669
  },
532
- "open_weights": false,
670
+ "open_weights": true,
533
671
  "limit": {
534
- "context": 262144,
535
- "output": 131072
672
+ "context": 131072,
673
+ "output": 65536
536
674
  },
537
675
  "cost": {
538
- "input": 0.15,
539
- "output": 0.6
676
+ "input": 0.22,
677
+ "output": 1.8
540
678
  }
541
679
  },
542
- "anthropic.claude-opus-4-7": {
543
- "id": "anthropic.claude-opus-4-7",
544
- "name": "Claude Opus 4.7",
680
+ "us.anthropic.claude-opus-4-7": {
681
+ "id": "us.anthropic.claude-opus-4-7",
682
+ "name": "Claude Opus 4.7 (US)",
683
+ "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
545
684
  "family": "claude-opus",
546
685
  "attachment": true,
547
686
  "reasoning": true,
687
+ "reasoning_options": [
688
+ {
689
+ "type": "effort",
690
+ "values": [
691
+ "low",
692
+ "medium",
693
+ "high",
694
+ "xhigh",
695
+ "max"
696
+ ]
697
+ }
698
+ ],
548
699
  "tool_call": true,
549
700
  "temperature": false,
550
701
  "knowledge": "2026-01-31",
@@ -572,18 +723,18 @@
572
723
  "cache_write": 6.25
573
724
  }
574
725
  },
575
- "zai.glm-4.7-flash": {
576
- "id": "zai.glm-4.7-flash",
577
- "name": "GLM-4.7-Flash",
578
- "family": "glm-flash",
726
+ "nvidia.nemotron-nano-9b-v2": {
727
+ "id": "nvidia.nemotron-nano-9b-v2",
728
+ "name": "NVIDIA Nemotron Nano 9B v2",
729
+ "description": "Compact Nemotron model for efficient reasoning and deployable AI agents",
730
+ "family": "nemotron",
579
731
  "attachment": false,
580
- "reasoning": true,
732
+ "reasoning": false,
581
733
  "tool_call": true,
582
734
  "structured_output": true,
583
735
  "temperature": true,
584
- "knowledge": "2025-04",
585
- "release_date": "2026-01-19",
586
- "last_updated": "2026-01-19",
736
+ "release_date": "2024-12-01",
737
+ "last_updated": "2024-12-01",
587
738
  "modalities": {
588
739
  "input": [
589
740
  "text"
@@ -592,60 +743,77 @@
592
743
  "text"
593
744
  ]
594
745
  },
595
- "open_weights": true,
746
+ "open_weights": false,
596
747
  "limit": {
597
- "context": 200000,
598
- "output": 131072
748
+ "context": 128000,
749
+ "output": 4096
599
750
  },
600
751
  "cost": {
601
- "input": 0.07,
602
- "output": 0.4
752
+ "input": 0.06,
753
+ "output": 0.23
603
754
  }
604
755
  },
605
- "google.gemma-3-12b-it": {
606
- "id": "google.gemma-3-12b-it",
607
- "name": "Google Gemma 3 12B",
608
- "family": "gemma",
756
+ "qwen.qwen3-32b-v1:0": {
757
+ "id": "qwen.qwen3-32b-v1:0",
758
+ "name": "Qwen3 32B (dense)",
759
+ "description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
760
+ "family": "qwen",
609
761
  "attachment": false,
610
- "reasoning": false,
611
- "tool_call": false,
762
+ "reasoning": true,
763
+ "reasoning_options": [],
764
+ "tool_call": true,
612
765
  "structured_output": true,
613
766
  "temperature": true,
614
- "knowledge": "2024-12",
615
- "release_date": "2024-12-01",
616
- "last_updated": "2024-12-01",
767
+ "knowledge": "2024-04",
768
+ "release_date": "2025-09-18",
769
+ "last_updated": "2025-09-18",
617
770
  "modalities": {
618
771
  "input": [
619
- "text",
620
- "image"
772
+ "text"
621
773
  ],
622
774
  "output": [
623
775
  "text"
624
776
  ]
625
777
  },
626
- "open_weights": false,
778
+ "open_weights": true,
627
779
  "limit": {
628
- "context": 131072,
629
- "output": 8192
780
+ "context": 16384,
781
+ "output": 16384
630
782
  },
631
783
  "cost": {
632
- "input": 0.049999999999999996,
633
- "output": 0.09999999999999999
784
+ "input": 0.15,
785
+ "output": 0.6
634
786
  }
635
787
  },
636
- "global.anthropic.claude-opus-4-5-20251101-v1:0": {
637
- "id": "global.anthropic.claude-opus-4-5-20251101-v1:0",
638
- "name": "Claude Opus 4.5 (Global)",
639
- "family": "claude-opus",
788
+ "jp.anthropic.claude-sonnet-4-6": {
789
+ "id": "jp.anthropic.claude-sonnet-4-6",
790
+ "name": "Claude Sonnet 4.6 (JP)",
791
+ "description": "Claude workhorse for coding agents, careful analysis, and production cost control",
792
+ "family": "claude-sonnet",
640
793
  "attachment": true,
641
794
  "reasoning": true,
795
+ "reasoning_options": [
796
+ {
797
+ "type": "effort",
798
+ "values": [
799
+ "low",
800
+ "medium",
801
+ "high",
802
+ "max"
803
+ ]
804
+ },
805
+ {
806
+ "type": "budget_tokens",
807
+ "min": 1024
808
+ }
809
+ ],
642
810
  "tool_call": true,
643
811
  "structured_output": true,
644
812
  "temperature": true,
645
- "knowledge": "2025-03-31",
646
- "release_date": "2025-11-24",
647
- "last_updated": "2025-08-01",
648
- "modalities": {
813
+ "knowledge": "2025-08-31",
814
+ "release_date": "2026-02-17",
815
+ "last_updated": "2026-03-13",
816
+ "modalities": {
649
817
  "input": [
650
818
  "text",
651
819
  "image",
@@ -657,33 +825,32 @@
657
825
  },
658
826
  "open_weights": false,
659
827
  "limit": {
660
- "context": 200000,
828
+ "context": 1000000,
661
829
  "output": 64000
662
830
  },
663
831
  "cost": {
664
- "input": 5,
665
- "output": 25,
666
- "cache_read": 0.5,
667
- "cache_write": 6.25
832
+ "input": 3,
833
+ "output": 15,
834
+ "cache_read": 0.3,
835
+ "cache_write": 3.75
668
836
  }
669
837
  },
670
- "au.anthropic.claude-sonnet-4-6": {
671
- "id": "au.anthropic.claude-sonnet-4-6",
672
- "name": "AU Anthropic Claude Sonnet 4.6",
673
- "family": "claude-sonnet",
674
- "attachment": true,
838
+ "deepseek.r1-v1:0": {
839
+ "id": "deepseek.r1-v1:0",
840
+ "name": "DeepSeek-R1",
841
+ "description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
842
+ "family": "deepseek-thinking",
843
+ "attachment": false,
675
844
  "reasoning": true,
845
+ "reasoning_options": [],
676
846
  "tool_call": true,
677
- "structured_output": true,
678
847
  "temperature": true,
679
- "knowledge": "2025-08",
680
- "release_date": "2026-02-17",
681
- "last_updated": "2026-02-17",
848
+ "knowledge": "2024-07",
849
+ "release_date": "2025-01-20",
850
+ "last_updated": "2025-05-29",
682
851
  "modalities": {
683
852
  "input": [
684
- "text",
685
- "image",
686
- "pdf"
853
+ "text"
687
854
  ],
688
855
  "output": [
689
856
  "text"
@@ -691,27 +858,26 @@
691
858
  },
692
859
  "open_weights": false,
693
860
  "limit": {
694
- "context": 1000000,
695
- "output": 128000
861
+ "context": 128000,
862
+ "output": 32768
696
863
  },
697
864
  "cost": {
698
- "input": 3.3,
699
- "output": 16.5,
700
- "cache_read": 0.33,
701
- "cache_write": 4.125
865
+ "input": 1.35,
866
+ "output": 5.4
702
867
  }
703
868
  },
704
- "qwen.qwen3-vl-235b-a22b": {
705
- "id": "qwen.qwen3-vl-235b-a22b",
706
- "name": "Qwen/Qwen3-VL-235B-A22B-Instruct",
707
- "family": "qwen",
708
- "attachment": true,
869
+ "mistral.mistral-large-3-675b-instruct": {
870
+ "id": "mistral.mistral-large-3-675b-instruct",
871
+ "name": "Mistral Large 3",
872
+ "description": "Flagship Mistral model for advanced reasoning, coding, and multilingual work",
873
+ "family": "mistral",
874
+ "attachment": false,
709
875
  "reasoning": false,
710
876
  "tool_call": true,
711
877
  "structured_output": true,
712
878
  "temperature": true,
713
- "release_date": "2025-10-04",
714
- "last_updated": "2025-11-25",
879
+ "release_date": "2025-12-02",
880
+ "last_updated": "2025-12-02",
715
881
  "modalities": {
716
882
  "input": [
717
883
  "text",
@@ -721,30 +887,33 @@
721
887
  "text"
722
888
  ]
723
889
  },
724
- "open_weights": false,
890
+ "open_weights": true,
725
891
  "limit": {
726
- "context": 262000,
727
- "output": 262000
892
+ "context": 256000,
893
+ "output": 8192
728
894
  },
729
895
  "cost": {
730
- "input": 0.3,
896
+ "input": 0.5,
731
897
  "output": 1.5
732
898
  }
733
899
  },
734
- "meta.llama3-3-70b-instruct-v1:0": {
735
- "id": "meta.llama3-3-70b-instruct-v1:0",
736
- "name": "Llama 3.3 70B Instruct",
737
- "family": "llama",
738
- "attachment": false,
900
+ "google.gemma-3-27b-it": {
901
+ "id": "google.gemma-3-27b-it",
902
+ "name": "Google Gemma 3 27B Instruct",
903
+ "description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
904
+ "family": "gemma",
905
+ "attachment": true,
739
906
  "reasoning": false,
740
907
  "tool_call": true,
908
+ "structured_output": true,
741
909
  "temperature": true,
742
- "knowledge": "2023-12",
743
- "release_date": "2024-12-06",
744
- "last_updated": "2024-12-06",
910
+ "knowledge": "2025-07",
911
+ "release_date": "2025-07-27",
912
+ "last_updated": "2025-07-27",
745
913
  "modalities": {
746
914
  "input": [
747
- "text"
915
+ "text",
916
+ "image"
748
917
  ],
749
918
  "output": [
750
919
  "text"
@@ -752,28 +921,47 @@
752
921
  },
753
922
  "open_weights": true,
754
923
  "limit": {
755
- "context": 128000,
756
- "output": 4096
924
+ "context": 202752,
925
+ "output": 8192
757
926
  },
758
927
  "cost": {
759
- "input": 0.72,
760
- "output": 0.72
928
+ "input": 0.12,
929
+ "output": 0.2
761
930
  }
762
931
  },
763
- "nvidia.nemotron-nano-9b-v2": {
764
- "id": "nvidia.nemotron-nano-9b-v2",
765
- "name": "NVIDIA Nemotron Nano 9B v2",
766
- "family": "nemotron",
767
- "attachment": false,
768
- "reasoning": false,
932
+ "anthropic.claude-sonnet-4-6": {
933
+ "id": "anthropic.claude-sonnet-4-6",
934
+ "name": "Claude Sonnet 4.6",
935
+ "description": "Claude workhorse for coding agents, careful analysis, and production cost control",
936
+ "family": "claude-sonnet",
937
+ "attachment": true,
938
+ "reasoning": true,
939
+ "reasoning_options": [
940
+ {
941
+ "type": "effort",
942
+ "values": [
943
+ "low",
944
+ "medium",
945
+ "high",
946
+ "max"
947
+ ]
948
+ },
949
+ {
950
+ "type": "budget_tokens",
951
+ "min": 1024
952
+ }
953
+ ],
769
954
  "tool_call": true,
770
955
  "structured_output": true,
771
956
  "temperature": true,
772
- "release_date": "2024-12-01",
773
- "last_updated": "2024-12-01",
957
+ "knowledge": "2025-08-31",
958
+ "release_date": "2026-02-17",
959
+ "last_updated": "2026-03-13",
774
960
  "modalities": {
775
961
  "input": [
776
- "text"
962
+ "text",
963
+ "image",
964
+ "pdf"
777
965
  ],
778
966
  "output": [
779
967
  "text"
@@ -781,30 +969,45 @@
781
969
  },
782
970
  "open_weights": false,
783
971
  "limit": {
784
- "context": 128000,
785
- "output": 4096
972
+ "context": 1000000,
973
+ "output": 64000
786
974
  },
787
975
  "cost": {
788
- "input": 0.06,
789
- "output": 0.23
976
+ "input": 3,
977
+ "output": 15,
978
+ "cache_read": 0.3,
979
+ "cache_write": 3.75
790
980
  }
791
981
  },
792
- "us.anthropic.claude-opus-4-7": {
793
- "id": "us.anthropic.claude-opus-4-7",
794
- "name": "Claude Opus 4.7 (US)",
795
- "family": "claude-opus",
796
- "attachment": true,
982
+ "amazon.nova-2-lite-v1:0": {
983
+ "id": "amazon.nova-2-lite-v1:0",
984
+ "name": "Nova 2 Lite",
985
+ "description": "Multimodal reasoning model for visual analysis, planning, and tool use",
986
+ "family": "nova",
987
+ "attachment": false,
797
988
  "reasoning": true,
989
+ "reasoning_options": [
990
+ {
991
+ "type": "toggle"
992
+ },
993
+ {
994
+ "type": "effort",
995
+ "values": [
996
+ "low",
997
+ "medium",
998
+ "high"
999
+ ]
1000
+ }
1001
+ ],
798
1002
  "tool_call": true,
799
- "temperature": false,
800
- "knowledge": "2026-01-31",
801
- "release_date": "2026-04-16",
802
- "last_updated": "2026-04-16",
1003
+ "temperature": true,
1004
+ "release_date": "2024-12-01",
1005
+ "last_updated": "2024-12-01",
803
1006
  "modalities": {
804
1007
  "input": [
805
1008
  "text",
806
1009
  "image",
807
- "pdf"
1010
+ "video"
808
1011
  ],
809
1012
  "output": [
810
1013
  "text"
@@ -812,31 +1015,29 @@
812
1015
  },
813
1016
  "open_weights": false,
814
1017
  "limit": {
815
- "context": 1000000,
816
- "output": 128000
1018
+ "context": 128000,
1019
+ "output": 4096
817
1020
  },
818
1021
  "cost": {
819
- "input": 5,
820
- "output": 25,
821
- "cache_read": 0.5,
822
- "cache_write": 6.25
1022
+ "input": 0.33,
1023
+ "output": 2.75
823
1024
  }
824
1025
  },
825
- "nvidia.nemotron-nano-12b-v2": {
826
- "id": "nvidia.nemotron-nano-12b-v2",
827
- "name": "NVIDIA Nemotron Nano 12B v2 VL BF16",
828
- "family": "nemotron",
1026
+ "openai.gpt-oss-safeguard-120b": {
1027
+ "id": "openai.gpt-oss-safeguard-120b",
1028
+ "name": "GPT OSS Safeguard 120B",
1029
+ "description": "Safety model for policy screening, moderation, and risk-aware routing workflows",
1030
+ "family": "gpt-oss",
829
1031
  "attachment": false,
830
1032
  "reasoning": false,
831
1033
  "tool_call": true,
832
1034
  "structured_output": true,
833
1035
  "temperature": true,
834
- "release_date": "2024-12-01",
835
- "last_updated": "2024-12-01",
1036
+ "release_date": "2025-10-29",
1037
+ "last_updated": "2025-10-29",
836
1038
  "modalities": {
837
1039
  "input": [
838
- "text",
839
- "image"
1040
+ "text"
840
1041
  ],
841
1042
  "output": [
842
1043
  "text"
@@ -845,30 +1046,28 @@
845
1046
  "open_weights": false,
846
1047
  "limit": {
847
1048
  "context": 128000,
848
- "output": 4096
1049
+ "output": 16384
849
1050
  },
850
1051
  "cost": {
851
- "input": 0.2,
1052
+ "input": 0.15,
852
1053
  "output": 0.6
853
1054
  }
854
1055
  },
855
- "anthropic.claude-opus-4-5-20251101-v1:0": {
856
- "id": "anthropic.claude-opus-4-5-20251101-v1:0",
857
- "name": "Claude Opus 4.5",
858
- "family": "claude-opus",
859
- "attachment": true,
860
- "reasoning": true,
1056
+ "mistral.ministral-3-8b-instruct": {
1057
+ "id": "mistral.ministral-3-8b-instruct",
1058
+ "name": "Ministral 3 8B",
1059
+ "description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
1060
+ "family": "ministral",
1061
+ "attachment": false,
1062
+ "reasoning": false,
861
1063
  "tool_call": true,
862
1064
  "structured_output": true,
863
1065
  "temperature": true,
864
- "knowledge": "2025-03-31",
865
- "release_date": "2025-11-24",
866
- "last_updated": "2025-08-01",
1066
+ "release_date": "2024-12-01",
1067
+ "last_updated": "2024-12-01",
867
1068
  "modalities": {
868
1069
  "input": [
869
- "text",
870
- "image",
871
- "pdf"
1070
+ "text"
872
1071
  ],
873
1072
  "output": [
874
1073
  "text"
@@ -876,87 +1075,142 @@
876
1075
  },
877
1076
  "open_weights": false,
878
1077
  "limit": {
879
- "context": 200000,
880
- "output": 64000
1078
+ "context": 128000,
1079
+ "output": 4096
881
1080
  },
882
1081
  "cost": {
883
- "input": 5,
884
- "output": 25,
885
- "cache_read": 0.5,
886
- "cache_write": 6.25
1082
+ "input": 0.15,
1083
+ "output": 0.15
887
1084
  }
888
1085
  },
889
- "deepseek.v3-v1:0": {
890
- "id": "deepseek.v3-v1:0",
891
- "name": "DeepSeek-V3.1",
892
- "family": "deepseek",
893
- "attachment": false,
1086
+ "eu.anthropic.claude-opus-4-6-v1": {
1087
+ "id": "eu.anthropic.claude-opus-4-6-v1",
1088
+ "name": "Claude Opus 4.6 (EU)",
1089
+ "description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
1090
+ "family": "claude-opus",
1091
+ "attachment": true,
894
1092
  "reasoning": true,
1093
+ "reasoning_options": [
1094
+ {
1095
+ "type": "effort",
1096
+ "values": [
1097
+ "low",
1098
+ "medium",
1099
+ "high",
1100
+ "max"
1101
+ ]
1102
+ },
1103
+ {
1104
+ "type": "budget_tokens",
1105
+ "min": 1024
1106
+ }
1107
+ ],
895
1108
  "tool_call": true,
896
1109
  "structured_output": true,
897
1110
  "temperature": true,
898
- "knowledge": "2024-07",
899
- "release_date": "2025-09-18",
900
- "last_updated": "2025-09-18",
1111
+ "knowledge": "2025-05-31",
1112
+ "release_date": "2026-02-05",
1113
+ "last_updated": "2026-03-13",
901
1114
  "modalities": {
902
1115
  "input": [
903
- "text"
1116
+ "text",
1117
+ "image",
1118
+ "pdf"
904
1119
  ],
905
1120
  "output": [
906
1121
  "text"
907
1122
  ]
908
1123
  },
909
- "open_weights": true,
1124
+ "open_weights": false,
910
1125
  "limit": {
911
- "context": 163840,
912
- "output": 81920
1126
+ "context": 1000000,
1127
+ "output": 128000
913
1128
  },
914
1129
  "cost": {
915
- "input": 0.58,
916
- "output": 1.68
1130
+ "input": 5.5,
1131
+ "output": 27.5,
1132
+ "cache_read": 0.55,
1133
+ "cache_write": 6.875
917
1134
  }
918
1135
  },
919
- "meta.llama4-maverick-17b-instruct-v1:0": {
920
- "id": "meta.llama4-maverick-17b-instruct-v1:0",
921
- "name": "Llama 4 Maverick 17B Instruct",
922
- "family": "llama",
1136
+ "au.anthropic.claude-opus-4-6-v1": {
1137
+ "id": "au.anthropic.claude-opus-4-6-v1",
1138
+ "name": "AU Anthropic Claude Opus 4.6",
1139
+ "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
1140
+ "family": "claude-opus",
923
1141
  "attachment": true,
924
- "reasoning": false,
1142
+ "reasoning": true,
1143
+ "reasoning_options": [
1144
+ {
1145
+ "type": "effort",
1146
+ "values": [
1147
+ "low",
1148
+ "medium",
1149
+ "high",
1150
+ "max"
1151
+ ]
1152
+ },
1153
+ {
1154
+ "type": "budget_tokens",
1155
+ "min": 1024
1156
+ }
1157
+ ],
925
1158
  "tool_call": true,
1159
+ "structured_output": true,
926
1160
  "temperature": true,
927
- "knowledge": "2024-08",
928
- "release_date": "2025-04-05",
929
- "last_updated": "2025-04-05",
1161
+ "knowledge": "2025-05",
1162
+ "release_date": "2026-02-05",
1163
+ "last_updated": "2026-02-05",
930
1164
  "modalities": {
931
1165
  "input": [
932
1166
  "text",
933
- "image"
1167
+ "image",
1168
+ "pdf"
934
1169
  ],
935
1170
  "output": [
936
1171
  "text"
937
1172
  ]
938
1173
  },
939
- "open_weights": true,
1174
+ "open_weights": false,
940
1175
  "limit": {
941
1176
  "context": 1000000,
942
- "output": 16384
1177
+ "output": 128000
943
1178
  },
944
1179
  "cost": {
945
- "input": 0.24,
946
- "output": 0.97
1180
+ "input": 16.5,
1181
+ "output": 82.5,
1182
+ "cache_read": 1.65,
1183
+ "cache_write": 20.625
947
1184
  }
948
1185
  },
949
- "eu.anthropic.claude-opus-4-7": {
950
- "id": "eu.anthropic.claude-opus-4-7",
951
- "name": "Claude Opus 4.7 (EU)",
952
- "family": "claude-opus",
1186
+ "jp.anthropic.claude-sonnet-5": {
1187
+ "id": "jp.anthropic.claude-sonnet-5",
1188
+ "name": "Claude Sonnet 5 (JP)",
1189
+ "description": "Everyday Claude agent model for coding, planning, browsing, and general work",
1190
+ "family": "claude-sonnet",
953
1191
  "attachment": true,
954
1192
  "reasoning": true,
1193
+ "reasoning_options": [
1194
+ {
1195
+ "type": "toggle"
1196
+ },
1197
+ {
1198
+ "type": "effort",
1199
+ "values": [
1200
+ "low",
1201
+ "medium",
1202
+ "high",
1203
+ "xhigh",
1204
+ "max"
1205
+ ]
1206
+ }
1207
+ ],
955
1208
  "tool_call": true,
1209
+ "structured_output": true,
956
1210
  "temperature": false,
957
1211
  "knowledge": "2026-01-31",
958
- "release_date": "2026-04-16",
959
- "last_updated": "2026-04-16",
1212
+ "release_date": "2026-06-30",
1213
+ "last_updated": "2026-06-30",
960
1214
  "modalities": {
961
1215
  "input": [
962
1216
  "text",
@@ -973,24 +1227,34 @@
973
1227
  "output": 128000
974
1228
  },
975
1229
  "cost": {
976
- "input": 5,
977
- "output": 25,
978
- "cache_read": 0.5,
979
- "cache_write": 6.25
1230
+ "input": 2,
1231
+ "output": 10,
1232
+ "cache_read": 0.2,
1233
+ "cache_write": 2.5
980
1234
  }
981
1235
  },
982
- "qwen.qwen3-32b-v1:0": {
983
- "id": "qwen.qwen3-32b-v1:0",
984
- "name": "Qwen3 32B (dense)",
985
- "family": "qwen",
1236
+ "openai.gpt-oss-120b": {
1237
+ "id": "openai.gpt-oss-120b",
1238
+ "name": "gpt-oss-120b",
1239
+ "description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
1240
+ "family": "gpt-oss",
986
1241
  "attachment": false,
987
1242
  "reasoning": true,
1243
+ "reasoning_options": [
1244
+ {
1245
+ "type": "effort",
1246
+ "values": [
1247
+ "low",
1248
+ "medium",
1249
+ "high"
1250
+ ]
1251
+ }
1252
+ ],
988
1253
  "tool_call": true,
989
1254
  "structured_output": true,
990
1255
  "temperature": true,
991
- "knowledge": "2024-04",
992
- "release_date": "2025-09-18",
993
- "last_updated": "2025-09-18",
1256
+ "release_date": "2025-08-05",
1257
+ "last_updated": "2025-08-05",
994
1258
  "modalities": {
995
1259
  "input": [
996
1260
  "text"
@@ -999,32 +1263,54 @@
999
1263
  "text"
1000
1264
  ]
1001
1265
  },
1002
- "open_weights": true,
1266
+ "open_weights": false,
1003
1267
  "limit": {
1004
- "context": 16384,
1268
+ "context": 128000,
1005
1269
  "output": 16384
1006
1270
  },
1271
+ "provider": {
1272
+ "npm": "@ai-sdk/amazon-bedrock/mantle",
1273
+ "api": "https://bedrock-mantle.${AWS_REGION}.api.aws/v1",
1274
+ "shape": "responses"
1275
+ },
1007
1276
  "cost": {
1008
1277
  "input": 0.15,
1009
1278
  "output": 0.6
1010
1279
  }
1011
1280
  },
1012
- "amazon.nova-lite-v1:0": {
1013
- "id": "amazon.nova-lite-v1:0",
1014
- "name": "Nova Lite",
1015
- "family": "nova-lite",
1281
+ "global.anthropic.claude-sonnet-5": {
1282
+ "id": "global.anthropic.claude-sonnet-5",
1283
+ "name": "Claude Sonnet 5 (Global)",
1284
+ "description": "Everyday Claude agent model for coding, planning, browsing, and general work",
1285
+ "family": "claude-sonnet",
1016
1286
  "attachment": true,
1017
- "reasoning": false,
1287
+ "reasoning": true,
1288
+ "reasoning_options": [
1289
+ {
1290
+ "type": "toggle"
1291
+ },
1292
+ {
1293
+ "type": "effort",
1294
+ "values": [
1295
+ "low",
1296
+ "medium",
1297
+ "high",
1298
+ "xhigh",
1299
+ "max"
1300
+ ]
1301
+ }
1302
+ ],
1018
1303
  "tool_call": true,
1019
- "temperature": true,
1020
- "knowledge": "2024-10",
1021
- "release_date": "2024-12-03",
1022
- "last_updated": "2024-12-03",
1304
+ "structured_output": true,
1305
+ "temperature": false,
1306
+ "knowledge": "2026-01-31",
1307
+ "release_date": "2026-06-30",
1308
+ "last_updated": "2026-06-30",
1023
1309
  "modalities": {
1024
1310
  "input": [
1025
1311
  "text",
1026
1312
  "image",
1027
- "video"
1313
+ "pdf"
1028
1314
  ],
1029
1315
  "output": [
1030
1316
  "text"
@@ -1032,92 +1318,122 @@
1032
1318
  },
1033
1319
  "open_weights": false,
1034
1320
  "limit": {
1035
- "context": 300000,
1036
- "output": 8192
1321
+ "context": 1000000,
1322
+ "output": 128000
1037
1323
  },
1038
1324
  "cost": {
1039
- "input": 0.06,
1040
- "output": 0.24,
1041
- "cache_read": 0.015
1325
+ "input": 2,
1326
+ "output": 10,
1327
+ "cache_read": 0.2,
1328
+ "cache_write": 2.5
1042
1329
  }
1043
1330
  },
1044
- "mistral.devstral-2-123b": {
1045
- "id": "mistral.devstral-2-123b",
1046
- "name": "Devstral 2 123B",
1047
- "family": "devstral",
1048
- "attachment": false,
1049
- "reasoning": false,
1331
+ "global.anthropic.claude-opus-4-8": {
1332
+ "id": "global.anthropic.claude-opus-4-8",
1333
+ "name": "Claude Opus 4.8 (Global)",
1334
+ "description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
1335
+ "family": "claude-opus",
1336
+ "attachment": true,
1337
+ "reasoning": true,
1338
+ "reasoning_options": [
1339
+ {
1340
+ "type": "effort",
1341
+ "values": [
1342
+ "low",
1343
+ "medium",
1344
+ "high",
1345
+ "xhigh",
1346
+ "max"
1347
+ ]
1348
+ }
1349
+ ],
1050
1350
  "tool_call": true,
1051
- "structured_output": true,
1052
- "temperature": true,
1053
- "release_date": "2026-02-17",
1054
- "last_updated": "2026-02-17",
1351
+ "temperature": false,
1352
+ "release_date": "2026-05-28",
1353
+ "last_updated": "2026-05-28",
1055
1354
  "modalities": {
1056
1355
  "input": [
1057
- "text"
1356
+ "text",
1357
+ "image",
1358
+ "pdf"
1058
1359
  ],
1059
1360
  "output": [
1060
1361
  "text"
1061
1362
  ]
1062
1363
  },
1063
- "open_weights": true,
1364
+ "open_weights": false,
1064
1365
  "limit": {
1065
- "context": 256000,
1066
- "output": 8192
1366
+ "context": 1000000,
1367
+ "output": 128000
1067
1368
  },
1068
1369
  "cost": {
1069
- "input": 0.4,
1070
- "output": 2
1370
+ "input": 5,
1371
+ "output": 25,
1372
+ "cache_read": 0.5,
1373
+ "cache_write": 6.25
1071
1374
  }
1072
1375
  },
1073
- "eu.anthropic.claude-opus-4-5-20251101-v1:0": {
1074
- "id": "eu.anthropic.claude-opus-4-5-20251101-v1:0",
1075
- "name": "Claude Opus 4.5 (EU)",
1076
- "family": "claude-opus",
1376
+ "us.meta.llama4-maverick-17b-instruct-v1:0": {
1377
+ "id": "us.meta.llama4-maverick-17b-instruct-v1:0",
1378
+ "name": "Llama 4 Maverick 17B Instruct (US)",
1379
+ "description": "Open multimodal Llama for strong reasoning with efficient everyday serving",
1380
+ "family": "llama",
1077
1381
  "attachment": true,
1078
- "reasoning": true,
1382
+ "reasoning": false,
1079
1383
  "tool_call": true,
1080
- "structured_output": true,
1081
1384
  "temperature": true,
1082
- "knowledge": "2025-03-31",
1083
- "release_date": "2025-11-24",
1084
- "last_updated": "2025-08-01",
1385
+ "knowledge": "2024-08",
1386
+ "release_date": "2025-04-05",
1387
+ "last_updated": "2025-04-05",
1085
1388
  "modalities": {
1086
1389
  "input": [
1087
1390
  "text",
1088
- "image",
1089
- "pdf"
1391
+ "image"
1090
1392
  ],
1091
1393
  "output": [
1092
1394
  "text"
1093
1395
  ]
1094
1396
  },
1095
- "open_weights": false,
1397
+ "open_weights": true,
1096
1398
  "limit": {
1097
- "context": 200000,
1098
- "output": 64000
1399
+ "context": 1000000,
1400
+ "output": 16384
1099
1401
  },
1100
1402
  "cost": {
1101
- "input": 5,
1102
- "output": 25,
1103
- "cache_read": 0.5,
1104
- "cache_write": 6.25
1403
+ "input": 0.24,
1404
+ "output": 0.97
1105
1405
  }
1106
1406
  },
1107
- "openai.gpt-oss-120b-1:0": {
1108
- "id": "openai.gpt-oss-120b-1:0",
1109
- "name": "gpt-oss-120b",
1110
- "family": "gpt-oss",
1111
- "attachment": false,
1112
- "reasoning": false,
1407
+ "openai.gpt-5.4": {
1408
+ "id": "openai.gpt-5.4",
1409
+ "name": "GPT-5.4",
1410
+ "description": "Agent-ready GPT for coding and computer-use workflows at a lower cost",
1411
+ "family": "gpt",
1412
+ "attachment": true,
1413
+ "reasoning": true,
1414
+ "reasoning_options": [
1415
+ {
1416
+ "type": "effort",
1417
+ "values": [
1418
+ "none",
1419
+ "low",
1420
+ "medium",
1421
+ "high",
1422
+ "xhigh"
1423
+ ]
1424
+ }
1425
+ ],
1113
1426
  "tool_call": true,
1114
1427
  "structured_output": true,
1115
- "temperature": true,
1116
- "release_date": "2025-08-05",
1117
- "last_updated": "2025-08-05",
1428
+ "temperature": false,
1429
+ "knowledge": "2025-08-31",
1430
+ "release_date": "2026-03-05",
1431
+ "last_updated": "2026-06-01",
1118
1432
  "modalities": {
1119
1433
  "input": [
1120
- "text"
1434
+ "text",
1435
+ "image",
1436
+ "pdf"
1121
1437
  ],
1122
1438
  "output": [
1123
1439
  "text"
@@ -1125,28 +1441,53 @@
1125
1441
  },
1126
1442
  "open_weights": false,
1127
1443
  "limit": {
1128
- "context": 128000,
1129
- "output": 16384
1444
+ "context": 272000,
1445
+ "output": 128000
1446
+ },
1447
+ "provider": {
1448
+ "npm": "@ai-sdk/amazon-bedrock/mantle",
1449
+ "api": "https://bedrock-mantle.${AWS_REGION}.api.aws/openai/v1",
1450
+ "shape": "responses"
1130
1451
  },
1131
1452
  "cost": {
1132
- "input": 0.15,
1133
- "output": 0.6
1453
+ "input": 2.75,
1454
+ "output": 16.5,
1455
+ "cache_read": 0.275
1134
1456
  }
1135
1457
  },
1136
- "deepseek.r1-v1:0": {
1137
- "id": "deepseek.r1-v1:0",
1138
- "name": "DeepSeek-R1",
1139
- "family": "deepseek-thinking",
1140
- "attachment": false,
1458
+ "anthropic.claude-sonnet-5": {
1459
+ "id": "anthropic.claude-sonnet-5",
1460
+ "name": "Claude Sonnet 5",
1461
+ "description": "Everyday Claude agent model for coding, planning, browsing, and general work",
1462
+ "family": "claude-sonnet",
1463
+ "attachment": true,
1141
1464
  "reasoning": true,
1465
+ "reasoning_options": [
1466
+ {
1467
+ "type": "toggle"
1468
+ },
1469
+ {
1470
+ "type": "effort",
1471
+ "values": [
1472
+ "low",
1473
+ "medium",
1474
+ "high",
1475
+ "xhigh",
1476
+ "max"
1477
+ ]
1478
+ }
1479
+ ],
1142
1480
  "tool_call": true,
1143
- "temperature": true,
1144
- "knowledge": "2024-07",
1145
- "release_date": "2025-01-20",
1146
- "last_updated": "2025-05-29",
1481
+ "structured_output": true,
1482
+ "temperature": false,
1483
+ "knowledge": "2026-01-31",
1484
+ "release_date": "2026-06-30",
1485
+ "last_updated": "2026-06-30",
1147
1486
  "modalities": {
1148
1487
  "input": [
1149
- "text"
1488
+ "text",
1489
+ "image",
1490
+ "pdf"
1150
1491
  ],
1151
1492
  "output": [
1152
1493
  "text"
@@ -1154,29 +1495,31 @@
1154
1495
  },
1155
1496
  "open_weights": false,
1156
1497
  "limit": {
1157
- "context": 128000,
1158
- "output": 32768
1498
+ "context": 1000000,
1499
+ "output": 128000
1159
1500
  },
1160
1501
  "cost": {
1161
- "input": 1.35,
1162
- "output": 5.4
1502
+ "input": 2,
1503
+ "output": 10,
1504
+ "cache_read": 0.2,
1505
+ "cache_write": 2.5
1163
1506
  }
1164
1507
  },
1165
- "mistral.mistral-large-3-675b-instruct": {
1166
- "id": "mistral.mistral-large-3-675b-instruct",
1167
- "name": "Mistral Large 3",
1168
- "family": "mistral",
1508
+ "mistral.devstral-2-123b": {
1509
+ "id": "mistral.devstral-2-123b",
1510
+ "name": "Devstral 2 123B",
1511
+ "description": "Mistral coding agent model for repository tasks and software engineering workflows",
1512
+ "family": "devstral",
1169
1513
  "attachment": false,
1170
1514
  "reasoning": false,
1171
1515
  "tool_call": true,
1172
1516
  "structured_output": true,
1173
1517
  "temperature": true,
1174
- "release_date": "2025-12-02",
1175
- "last_updated": "2025-12-02",
1518
+ "release_date": "2026-02-17",
1519
+ "last_updated": "2026-02-17",
1176
1520
  "modalities": {
1177
1521
  "input": [
1178
- "text",
1179
- "image"
1522
+ "text"
1180
1523
  ],
1181
1524
  "output": [
1182
1525
  "text"
@@ -1188,25 +1531,30 @@
1188
1531
  "output": 8192
1189
1532
  },
1190
1533
  "cost": {
1191
- "input": 0.5,
1192
- "output": 1.5
1534
+ "input": 0.4,
1535
+ "output": 2
1193
1536
  }
1194
1537
  },
1195
- "mistral.magistral-small-2509": {
1196
- "id": "mistral.magistral-small-2509",
1197
- "name": "Magistral Small 1.2",
1198
- "family": "magistral",
1538
+ "zai.glm-4.7": {
1539
+ "id": "zai.glm-4.7",
1540
+ "name": "GLM-4.7",
1541
+ "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
1542
+ "family": "glm",
1199
1543
  "attachment": false,
1200
1544
  "reasoning": true,
1545
+ "reasoning_options": [],
1201
1546
  "tool_call": true,
1547
+ "interleaved": {
1548
+ "field": "reasoning_content"
1549
+ },
1202
1550
  "structured_output": true,
1203
1551
  "temperature": true,
1204
- "release_date": "2025-12-02",
1205
- "last_updated": "2025-12-02",
1552
+ "knowledge": "2025-04",
1553
+ "release_date": "2025-12-22",
1554
+ "last_updated": "2025-12-22",
1206
1555
  "modalities": {
1207
1556
  "input": [
1208
- "text",
1209
- "image"
1557
+ "text"
1210
1558
  ],
1211
1559
  "output": [
1212
1560
  "text"
@@ -1214,28 +1562,43 @@
1214
1562
  },
1215
1563
  "open_weights": true,
1216
1564
  "limit": {
1217
- "context": 128000,
1218
- "output": 40000
1565
+ "context": 204800,
1566
+ "output": 131072
1219
1567
  },
1220
1568
  "cost": {
1221
- "input": 0.5,
1222
- "output": 1.5
1569
+ "input": 0.6,
1570
+ "output": 2.2
1223
1571
  }
1224
1572
  },
1225
- "mistral.pixtral-large-2502-v1:0": {
1226
- "id": "mistral.pixtral-large-2502-v1:0",
1227
- "name": "Pixtral Large (25.02)",
1228
- "family": "mistral",
1229
- "attachment": false,
1230
- "reasoning": false,
1573
+ "anthropic.claude-fable-5": {
1574
+ "id": "anthropic.claude-fable-5",
1575
+ "name": "Claude Fable 5",
1576
+ "description": "Claude model for creative writing, analysis, and controlled agent workflows",
1577
+ "family": "claude-fable",
1578
+ "attachment": true,
1579
+ "reasoning": true,
1580
+ "reasoning_options": [
1581
+ {
1582
+ "type": "effort",
1583
+ "values": [
1584
+ "low",
1585
+ "medium",
1586
+ "high",
1587
+ "xhigh",
1588
+ "max"
1589
+ ]
1590
+ }
1591
+ ],
1231
1592
  "tool_call": true,
1232
- "temperature": true,
1233
- "release_date": "2025-04-08",
1234
- "last_updated": "2025-04-08",
1593
+ "temperature": false,
1594
+ "knowledge": "2026-01-31",
1595
+ "release_date": "2026-06-09",
1596
+ "last_updated": "2026-06-09",
1235
1597
  "modalities": {
1236
1598
  "input": [
1237
1599
  "text",
1238
- "image"
1600
+ "image",
1601
+ "pdf"
1239
1602
  ],
1240
1603
  "output": [
1241
1604
  "text"
@@ -1243,28 +1606,30 @@
1243
1606
  },
1244
1607
  "open_weights": false,
1245
1608
  "limit": {
1246
- "context": 128000,
1247
- "output": 8192
1609
+ "context": 1000000,
1610
+ "output": 128000
1248
1611
  },
1249
1612
  "cost": {
1250
- "input": 2,
1251
- "output": 6
1613
+ "input": 10,
1614
+ "output": 50,
1615
+ "cache_read": 1,
1616
+ "cache_write": 12.5
1252
1617
  }
1253
1618
  },
1254
- "mistral.voxtral-mini-3b-2507": {
1255
- "id": "mistral.voxtral-mini-3b-2507",
1256
- "name": "Voxtral Mini 3B 2507",
1257
- "family": "mistral",
1619
+ "writer.palmyra-x4-v1:0": {
1620
+ "id": "writer.palmyra-x4-v1:0",
1621
+ "name": "Palmyra X4",
1622
+ "description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use",
1623
+ "family": "palmyra",
1258
1624
  "attachment": false,
1259
- "reasoning": false,
1625
+ "reasoning": true,
1626
+ "reasoning_options": [],
1260
1627
  "tool_call": true,
1261
- "structured_output": true,
1262
1628
  "temperature": true,
1263
- "release_date": "2024-12-01",
1264
- "last_updated": "2024-12-01",
1629
+ "release_date": "2025-04-28",
1630
+ "last_updated": "2025-04-28",
1265
1631
  "modalities": {
1266
1632
  "input": [
1267
- "audio",
1268
1633
  "text"
1269
1634
  ],
1270
1635
  "output": [
@@ -1273,27 +1638,749 @@
1273
1638
  },
1274
1639
  "open_weights": false,
1275
1640
  "limit": {
1276
- "context": 128000,
1641
+ "context": 122880,
1642
+ "output": 8192
1643
+ },
1644
+ "cost": {
1645
+ "input": 2.5,
1646
+ "output": 10
1647
+ }
1648
+ },
1649
+ "mistral.magistral-small-2509": {
1650
+ "id": "mistral.magistral-small-2509",
1651
+ "name": "Magistral Small 1.2",
1652
+ "description": "Mistral reasoning model for transparent analysis, math, and complex decisions",
1653
+ "family": "magistral",
1654
+ "attachment": false,
1655
+ "reasoning": true,
1656
+ "reasoning_options": [],
1657
+ "tool_call": true,
1658
+ "structured_output": true,
1659
+ "temperature": true,
1660
+ "release_date": "2025-12-02",
1661
+ "last_updated": "2025-12-02",
1662
+ "modalities": {
1663
+ "input": [
1664
+ "text",
1665
+ "image"
1666
+ ],
1667
+ "output": [
1668
+ "text"
1669
+ ]
1670
+ },
1671
+ "open_weights": true,
1672
+ "limit": {
1673
+ "context": 128000,
1674
+ "output": 40000
1675
+ },
1676
+ "cost": {
1677
+ "input": 0.5,
1678
+ "output": 1.5
1679
+ }
1680
+ },
1681
+ "qwen.qwen3-coder-480b-a35b-v1:0": {
1682
+ "id": "qwen.qwen3-coder-480b-a35b-v1:0",
1683
+ "name": "Qwen3 Coder 480B A35B Instruct",
1684
+ "description": "Qwen coding model for software agents, repository edits, and code reasoning",
1685
+ "family": "qwen",
1686
+ "attachment": false,
1687
+ "reasoning": false,
1688
+ "tool_call": true,
1689
+ "structured_output": true,
1690
+ "temperature": true,
1691
+ "knowledge": "2024-04",
1692
+ "release_date": "2025-09-18",
1693
+ "last_updated": "2025-09-18",
1694
+ "modalities": {
1695
+ "input": [
1696
+ "text"
1697
+ ],
1698
+ "output": [
1699
+ "text"
1700
+ ]
1701
+ },
1702
+ "open_weights": true,
1703
+ "limit": {
1704
+ "context": 131072,
1705
+ "output": 65536
1706
+ },
1707
+ "cost": {
1708
+ "input": 0.22,
1709
+ "output": 1.8
1710
+ }
1711
+ },
1712
+ "amazon.nova-micro-v1:0": {
1713
+ "id": "amazon.nova-micro-v1:0",
1714
+ "name": "Nova Micro",
1715
+ "description": "Efficient model for low-latency assistance, extraction, and routine automation",
1716
+ "family": "nova-micro",
1717
+ "attachment": false,
1718
+ "reasoning": false,
1719
+ "tool_call": true,
1720
+ "temperature": true,
1721
+ "knowledge": "2024-10",
1722
+ "release_date": "2024-12-03",
1723
+ "last_updated": "2024-12-03",
1724
+ "modalities": {
1725
+ "input": [
1726
+ "text"
1727
+ ],
1728
+ "output": [
1729
+ "text"
1730
+ ]
1731
+ },
1732
+ "open_weights": false,
1733
+ "limit": {
1734
+ "context": 128000,
1735
+ "output": 8192
1736
+ },
1737
+ "cost": {
1738
+ "input": 0.035,
1739
+ "output": 0.14,
1740
+ "cache_read": 0.00875
1741
+ }
1742
+ },
1743
+ "mistral.pixtral-large-2502-v1:0": {
1744
+ "id": "mistral.pixtral-large-2502-v1:0",
1745
+ "name": "Pixtral Large (25.02)",
1746
+ "description": "Mistral vision-language model for image understanding and multimodal chat",
1747
+ "family": "mistral",
1748
+ "attachment": false,
1749
+ "reasoning": false,
1750
+ "tool_call": true,
1751
+ "temperature": true,
1752
+ "release_date": "2025-04-08",
1753
+ "last_updated": "2025-04-08",
1754
+ "modalities": {
1755
+ "input": [
1756
+ "text",
1757
+ "image"
1758
+ ],
1759
+ "output": [
1760
+ "text"
1761
+ ]
1762
+ },
1763
+ "open_weights": false,
1764
+ "limit": {
1765
+ "context": 128000,
1766
+ "output": 8192
1767
+ },
1768
+ "cost": {
1769
+ "input": 2,
1770
+ "output": 6
1771
+ }
1772
+ },
1773
+ "us.anthropic.claude-opus-4-6-v1": {
1774
+ "id": "us.anthropic.claude-opus-4-6-v1",
1775
+ "name": "Claude Opus 4.6 (US)",
1776
+ "description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
1777
+ "family": "claude-opus",
1778
+ "attachment": true,
1779
+ "reasoning": true,
1780
+ "reasoning_options": [
1781
+ {
1782
+ "type": "effort",
1783
+ "values": [
1784
+ "low",
1785
+ "medium",
1786
+ "high",
1787
+ "max"
1788
+ ]
1789
+ },
1790
+ {
1791
+ "type": "budget_tokens",
1792
+ "min": 1024
1793
+ }
1794
+ ],
1795
+ "tool_call": true,
1796
+ "structured_output": true,
1797
+ "temperature": true,
1798
+ "knowledge": "2025-05-31",
1799
+ "release_date": "2026-02-05",
1800
+ "last_updated": "2026-03-13",
1801
+ "modalities": {
1802
+ "input": [
1803
+ "text",
1804
+ "image",
1805
+ "pdf"
1806
+ ],
1807
+ "output": [
1808
+ "text"
1809
+ ]
1810
+ },
1811
+ "open_weights": false,
1812
+ "limit": {
1813
+ "context": 1000000,
1814
+ "output": 128000
1815
+ },
1816
+ "cost": {
1817
+ "input": 5,
1818
+ "output": 25,
1819
+ "cache_read": 0.5,
1820
+ "cache_write": 6.25
1821
+ }
1822
+ },
1823
+ "jp.anthropic.claude-opus-4-7": {
1824
+ "id": "jp.anthropic.claude-opus-4-7",
1825
+ "name": "Claude Opus 4.7 (JP)",
1826
+ "description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
1827
+ "family": "claude-opus",
1828
+ "attachment": true,
1829
+ "reasoning": true,
1830
+ "reasoning_options": [
1831
+ {
1832
+ "type": "effort",
1833
+ "values": [
1834
+ "low",
1835
+ "medium",
1836
+ "high",
1837
+ "xhigh",
1838
+ "max"
1839
+ ]
1840
+ }
1841
+ ],
1842
+ "tool_call": true,
1843
+ "temperature": false,
1844
+ "knowledge": "2026-01-31",
1845
+ "release_date": "2026-04-16",
1846
+ "last_updated": "2026-04-16",
1847
+ "modalities": {
1848
+ "input": [
1849
+ "text",
1850
+ "image",
1851
+ "pdf"
1852
+ ],
1853
+ "output": [
1854
+ "text"
1855
+ ]
1856
+ },
1857
+ "open_weights": false,
1858
+ "limit": {
1859
+ "context": 1000000,
1860
+ "output": 128000
1861
+ },
1862
+ "cost": {
1863
+ "input": 5,
1864
+ "output": 25,
1865
+ "cache_read": 0.5,
1866
+ "cache_write": 6.25
1867
+ }
1868
+ },
1869
+ "au.anthropic.claude-sonnet-4-5-20250929-v1:0": {
1870
+ "id": "au.anthropic.claude-sonnet-4-5-20250929-v1:0",
1871
+ "name": "Claude Sonnet 4.5 (AU)",
1872
+ "description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
1873
+ "family": "claude-sonnet",
1874
+ "attachment": true,
1875
+ "reasoning": true,
1876
+ "reasoning_options": [
1877
+ {
1878
+ "type": "budget_tokens",
1879
+ "min": 1024
1880
+ }
1881
+ ],
1882
+ "tool_call": true,
1883
+ "structured_output": true,
1884
+ "temperature": true,
1885
+ "knowledge": "2025-07-31",
1886
+ "release_date": "2025-09-29",
1887
+ "last_updated": "2025-09-29",
1888
+ "modalities": {
1889
+ "input": [
1890
+ "text",
1891
+ "image",
1892
+ "pdf"
1893
+ ],
1894
+ "output": [
1895
+ "text"
1896
+ ]
1897
+ },
1898
+ "open_weights": false,
1899
+ "limit": {
1900
+ "context": 200000,
1901
+ "output": 64000
1902
+ },
1903
+ "cost": {
1904
+ "input": 3,
1905
+ "output": 15,
1906
+ "cache_read": 0.3,
1907
+ "cache_write": 3.75
1908
+ }
1909
+ },
1910
+ "deepseek.v3-v1:0": {
1911
+ "id": "deepseek.v3-v1:0",
1912
+ "name": "DeepSeek-V3.1",
1913
+ "description": "DeepSeek chat model for instruction following, coding, and analysis",
1914
+ "family": "deepseek",
1915
+ "attachment": false,
1916
+ "reasoning": true,
1917
+ "reasoning_options": [],
1918
+ "tool_call": true,
1919
+ "structured_output": true,
1920
+ "temperature": true,
1921
+ "knowledge": "2024-07",
1922
+ "release_date": "2025-09-18",
1923
+ "last_updated": "2025-09-18",
1924
+ "modalities": {
1925
+ "input": [
1926
+ "text"
1927
+ ],
1928
+ "output": [
1929
+ "text"
1930
+ ]
1931
+ },
1932
+ "open_weights": true,
1933
+ "limit": {
1934
+ "context": 163840,
1935
+ "output": 81920
1936
+ },
1937
+ "cost": {
1938
+ "input": 0.58,
1939
+ "output": 1.68
1940
+ }
1941
+ },
1942
+ "anthropic.claude-opus-4-1-20250805-v1:0": {
1943
+ "id": "anthropic.claude-opus-4-1-20250805-v1:0",
1944
+ "name": "Claude Opus 4.1",
1945
+ "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
1946
+ "family": "claude-opus",
1947
+ "attachment": true,
1948
+ "reasoning": true,
1949
+ "reasoning_options": [
1950
+ {
1951
+ "type": "budget_tokens",
1952
+ "min": 1024
1953
+ }
1954
+ ],
1955
+ "tool_call": true,
1956
+ "temperature": true,
1957
+ "knowledge": "2025-03-31",
1958
+ "release_date": "2025-08-05",
1959
+ "last_updated": "2025-08-05",
1960
+ "modalities": {
1961
+ "input": [
1962
+ "text",
1963
+ "image",
1964
+ "pdf"
1965
+ ],
1966
+ "output": [
1967
+ "text"
1968
+ ]
1969
+ },
1970
+ "open_weights": false,
1971
+ "limit": {
1972
+ "context": 200000,
1973
+ "output": 32000
1974
+ },
1975
+ "cost": {
1976
+ "input": 15,
1977
+ "output": 75,
1978
+ "cache_read": 1.5,
1979
+ "cache_write": 18.75
1980
+ }
1981
+ },
1982
+ "jp.anthropic.claude-sonnet-4-5-20250929-v1:0": {
1983
+ "id": "jp.anthropic.claude-sonnet-4-5-20250929-v1:0",
1984
+ "name": "Claude Sonnet 4.5 (JP)",
1985
+ "description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
1986
+ "family": "claude-sonnet",
1987
+ "attachment": true,
1988
+ "reasoning": true,
1989
+ "reasoning_options": [
1990
+ {
1991
+ "type": "budget_tokens",
1992
+ "min": 1024
1993
+ }
1994
+ ],
1995
+ "tool_call": true,
1996
+ "structured_output": true,
1997
+ "temperature": true,
1998
+ "knowledge": "2025-07-31",
1999
+ "release_date": "2025-09-29",
2000
+ "last_updated": "2025-09-29",
2001
+ "modalities": {
2002
+ "input": [
2003
+ "text",
2004
+ "image",
2005
+ "pdf"
2006
+ ],
2007
+ "output": [
2008
+ "text"
2009
+ ]
2010
+ },
2011
+ "open_weights": false,
2012
+ "limit": {
2013
+ "context": 200000,
2014
+ "output": 64000
2015
+ },
2016
+ "cost": {
2017
+ "input": 3,
2018
+ "output": 15,
2019
+ "cache_read": 0.3,
2020
+ "cache_write": 3.75
2021
+ }
2022
+ },
2023
+ "google.gemma-3-4b-it": {
2024
+ "id": "google.gemma-3-4b-it",
2025
+ "name": "Gemma 3 4B IT",
2026
+ "description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
2027
+ "family": "gemma",
2028
+ "attachment": false,
2029
+ "reasoning": false,
2030
+ "tool_call": true,
2031
+ "temperature": true,
2032
+ "release_date": "2024-12-01",
2033
+ "last_updated": "2024-12-01",
2034
+ "modalities": {
2035
+ "input": [
2036
+ "text",
2037
+ "image"
2038
+ ],
2039
+ "output": [
2040
+ "text"
2041
+ ]
2042
+ },
2043
+ "open_weights": false,
2044
+ "limit": {
2045
+ "context": 128000,
2046
+ "output": 4096
2047
+ },
2048
+ "cost": {
2049
+ "input": 0.04,
2050
+ "output": 0.08
2051
+ }
2052
+ },
2053
+ "eu.anthropic.claude-sonnet-4-5-20250929-v1:0": {
2054
+ "id": "eu.anthropic.claude-sonnet-4-5-20250929-v1:0",
2055
+ "name": "Claude Sonnet 4.5 (EU)",
2056
+ "description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
2057
+ "family": "claude-sonnet",
2058
+ "attachment": true,
2059
+ "reasoning": true,
2060
+ "reasoning_options": [
2061
+ {
2062
+ "type": "budget_tokens",
2063
+ "min": 1024
2064
+ }
2065
+ ],
2066
+ "tool_call": true,
2067
+ "structured_output": true,
2068
+ "temperature": true,
2069
+ "knowledge": "2025-07-31",
2070
+ "release_date": "2025-09-29",
2071
+ "last_updated": "2025-09-29",
2072
+ "modalities": {
2073
+ "input": [
2074
+ "text",
2075
+ "image",
2076
+ "pdf"
2077
+ ],
2078
+ "output": [
2079
+ "text"
2080
+ ]
2081
+ },
2082
+ "open_weights": false,
2083
+ "limit": {
2084
+ "context": 200000,
2085
+ "output": 64000
2086
+ },
2087
+ "cost": {
2088
+ "input": 3.3,
2089
+ "output": 16.5,
2090
+ "cache_read": 0.33,
2091
+ "cache_write": 4.125
2092
+ }
2093
+ },
2094
+ "qwen.qwen3-vl-235b-a22b": {
2095
+ "id": "qwen.qwen3-vl-235b-a22b",
2096
+ "name": "Qwen/Qwen3-VL-235B-A22B-Instruct",
2097
+ "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
2098
+ "family": "qwen",
2099
+ "attachment": true,
2100
+ "reasoning": false,
2101
+ "tool_call": true,
2102
+ "structured_output": true,
2103
+ "temperature": true,
2104
+ "release_date": "2025-10-04",
2105
+ "last_updated": "2025-11-25",
2106
+ "modalities": {
2107
+ "input": [
2108
+ "text",
2109
+ "image"
2110
+ ],
2111
+ "output": [
2112
+ "text"
2113
+ ]
2114
+ },
2115
+ "open_weights": false,
2116
+ "limit": {
2117
+ "context": 262000,
2118
+ "output": 262000
2119
+ },
2120
+ "cost": {
2121
+ "input": 0.3,
2122
+ "output": 1.5
2123
+ }
2124
+ },
2125
+ "writer.palmyra-x5-v1:0": {
2126
+ "id": "writer.palmyra-x5-v1:0",
2127
+ "name": "Palmyra X5",
2128
+ "description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use",
2129
+ "family": "palmyra",
2130
+ "attachment": false,
2131
+ "reasoning": true,
2132
+ "reasoning_options": [],
2133
+ "tool_call": true,
2134
+ "temperature": true,
2135
+ "release_date": "2025-04-28",
2136
+ "last_updated": "2025-04-28",
2137
+ "modalities": {
2138
+ "input": [
2139
+ "text"
2140
+ ],
2141
+ "output": [
2142
+ "text"
2143
+ ]
2144
+ },
2145
+ "open_weights": false,
2146
+ "limit": {
2147
+ "context": 1040000,
2148
+ "output": 8192
2149
+ },
2150
+ "cost": {
2151
+ "input": 0.6,
2152
+ "output": 6
2153
+ }
2154
+ },
2155
+ "us.anthropic.claude-sonnet-4-6": {
2156
+ "id": "us.anthropic.claude-sonnet-4-6",
2157
+ "name": "Claude Sonnet 4.6 (US)",
2158
+ "description": "Claude workhorse for coding agents, careful analysis, and production cost control",
2159
+ "family": "claude-sonnet",
2160
+ "attachment": true,
2161
+ "reasoning": true,
2162
+ "reasoning_options": [
2163
+ {
2164
+ "type": "effort",
2165
+ "values": [
2166
+ "low",
2167
+ "medium",
2168
+ "high",
2169
+ "max"
2170
+ ]
2171
+ },
2172
+ {
2173
+ "type": "budget_tokens",
2174
+ "min": 1024
2175
+ }
2176
+ ],
2177
+ "tool_call": true,
2178
+ "structured_output": true,
2179
+ "temperature": true,
2180
+ "knowledge": "2025-08-31",
2181
+ "release_date": "2026-02-17",
2182
+ "last_updated": "2026-03-13",
2183
+ "modalities": {
2184
+ "input": [
2185
+ "text",
2186
+ "image",
2187
+ "pdf"
2188
+ ],
2189
+ "output": [
2190
+ "text"
2191
+ ]
2192
+ },
2193
+ "open_weights": false,
2194
+ "limit": {
2195
+ "context": 1000000,
2196
+ "output": 64000
2197
+ },
2198
+ "cost": {
2199
+ "input": 3,
2200
+ "output": 15,
2201
+ "cache_read": 0.3,
2202
+ "cache_write": 3.75
2203
+ }
2204
+ },
2205
+ "au.anthropic.claude-haiku-4-5-20251001-v1:0": {
2206
+ "id": "au.anthropic.claude-haiku-4-5-20251001-v1:0",
2207
+ "name": "Claude Haiku 4.5 (AU)",
2208
+ "description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
2209
+ "family": "claude-haiku",
2210
+ "attachment": true,
2211
+ "reasoning": true,
2212
+ "reasoning_options": [
2213
+ {
2214
+ "type": "budget_tokens",
2215
+ "min": 1024
2216
+ }
2217
+ ],
2218
+ "tool_call": true,
2219
+ "structured_output": true,
2220
+ "temperature": true,
2221
+ "knowledge": "2025-02-28",
2222
+ "release_date": "2025-10-15",
2223
+ "last_updated": "2025-10-15",
2224
+ "modalities": {
2225
+ "input": [
2226
+ "text",
2227
+ "image",
2228
+ "pdf"
2229
+ ],
2230
+ "output": [
2231
+ "text"
2232
+ ]
2233
+ },
2234
+ "open_weights": false,
2235
+ "limit": {
2236
+ "context": 200000,
2237
+ "output": 64000
2238
+ },
2239
+ "cost": {
2240
+ "input": 1,
2241
+ "output": 5,
2242
+ "cache_read": 0.1,
2243
+ "cache_write": 1.25
2244
+ }
2245
+ },
2246
+ "meta.llama3-3-70b-instruct-v1:0": {
2247
+ "id": "meta.llama3-3-70b-instruct-v1:0",
2248
+ "name": "Llama 3.3 70B Instruct",
2249
+ "description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
2250
+ "family": "llama",
2251
+ "attachment": false,
2252
+ "reasoning": false,
2253
+ "tool_call": true,
2254
+ "temperature": true,
2255
+ "knowledge": "2023-12",
2256
+ "release_date": "2024-12-06",
2257
+ "last_updated": "2024-12-06",
2258
+ "modalities": {
2259
+ "input": [
2260
+ "text"
2261
+ ],
2262
+ "output": [
2263
+ "text"
2264
+ ]
2265
+ },
2266
+ "open_weights": true,
2267
+ "limit": {
2268
+ "context": 128000,
1277
2269
  "output": 4096
1278
2270
  },
1279
2271
  "cost": {
1280
- "input": 0.04,
1281
- "output": 0.04
2272
+ "input": 0.72,
2273
+ "output": 0.72
1282
2274
  }
1283
2275
  },
1284
- "writer.palmyra-x5-v1:0": {
1285
- "id": "writer.palmyra-x5-v1:0",
1286
- "name": "Palmyra X5",
1287
- "family": "palmyra",
2276
+ "zai.glm-5": {
2277
+ "id": "zai.glm-5",
2278
+ "name": "GLM-5",
2279
+ "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
2280
+ "family": "glm",
1288
2281
  "attachment": false,
1289
2282
  "reasoning": true,
2283
+ "reasoning_options": [],
1290
2284
  "tool_call": true,
2285
+ "interleaved": {
2286
+ "field": "reasoning_content"
2287
+ },
2288
+ "structured_output": true,
1291
2289
  "temperature": true,
1292
- "release_date": "2025-04-28",
1293
- "last_updated": "2025-04-28",
2290
+ "release_date": "2026-03-18",
2291
+ "last_updated": "2026-03-18",
2292
+ "modalities": {
2293
+ "input": [
2294
+ "text"
2295
+ ],
2296
+ "output": [
2297
+ "text"
2298
+ ]
2299
+ },
2300
+ "open_weights": true,
2301
+ "limit": {
2302
+ "context": 202752,
2303
+ "output": 101376
2304
+ },
2305
+ "cost": {
2306
+ "input": 1,
2307
+ "output": 3.2
2308
+ }
2309
+ },
2310
+ "us.anthropic.claude-opus-4-8": {
2311
+ "id": "us.anthropic.claude-opus-4-8",
2312
+ "name": "Claude Opus 4.8 (US)",
2313
+ "description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
2314
+ "family": "claude-opus",
2315
+ "attachment": true,
2316
+ "reasoning": true,
2317
+ "reasoning_options": [
2318
+ {
2319
+ "type": "effort",
2320
+ "values": [
2321
+ "low",
2322
+ "medium",
2323
+ "high",
2324
+ "xhigh",
2325
+ "max"
2326
+ ]
2327
+ }
2328
+ ],
2329
+ "tool_call": true,
2330
+ "temperature": false,
2331
+ "release_date": "2026-05-28",
2332
+ "last_updated": "2026-05-28",
1294
2333
  "modalities": {
1295
2334
  "input": [
2335
+ "text",
2336
+ "image",
2337
+ "pdf"
2338
+ ],
2339
+ "output": [
1296
2340
  "text"
2341
+ ]
2342
+ },
2343
+ "open_weights": false,
2344
+ "limit": {
2345
+ "context": 1000000,
2346
+ "output": 128000
2347
+ },
2348
+ "cost": {
2349
+ "input": 5,
2350
+ "output": 25,
2351
+ "cache_read": 0.5,
2352
+ "cache_write": 6.25
2353
+ }
2354
+ },
2355
+ "global.anthropic.claude-opus-4-7": {
2356
+ "id": "global.anthropic.claude-opus-4-7",
2357
+ "name": "Claude Opus 4.7 (Global)",
2358
+ "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
2359
+ "family": "claude-opus",
2360
+ "attachment": true,
2361
+ "reasoning": true,
2362
+ "reasoning_options": [
2363
+ {
2364
+ "type": "effort",
2365
+ "values": [
2366
+ "low",
2367
+ "medium",
2368
+ "high",
2369
+ "xhigh",
2370
+ "max"
2371
+ ]
2372
+ }
2373
+ ],
2374
+ "tool_call": true,
2375
+ "temperature": false,
2376
+ "knowledge": "2026-01-31",
2377
+ "release_date": "2026-04-16",
2378
+ "last_updated": "2026-04-16",
2379
+ "modalities": {
2380
+ "input": [
2381
+ "text",
2382
+ "image",
2383
+ "pdf"
1297
2384
  ],
1298
2385
  "output": [
1299
2386
  "text"
@@ -1301,56 +2388,199 @@
1301
2388
  },
1302
2389
  "open_weights": false,
1303
2390
  "limit": {
1304
- "context": 1040000,
2391
+ "context": 1000000,
2392
+ "output": 128000
2393
+ },
2394
+ "cost": {
2395
+ "input": 5,
2396
+ "output": 25,
2397
+ "cache_read": 0.5,
2398
+ "cache_write": 6.25
2399
+ }
2400
+ },
2401
+ "us.anthropic.claude-fable-5": {
2402
+ "id": "us.anthropic.claude-fable-5",
2403
+ "name": "Claude Fable 5 (US)",
2404
+ "description": "Claude model for creative writing, analysis, and controlled agent workflows",
2405
+ "family": "claude-fable",
2406
+ "attachment": true,
2407
+ "reasoning": true,
2408
+ "reasoning_options": [
2409
+ {
2410
+ "type": "effort",
2411
+ "values": [
2412
+ "low",
2413
+ "medium",
2414
+ "high",
2415
+ "xhigh",
2416
+ "max"
2417
+ ]
2418
+ }
2419
+ ],
2420
+ "tool_call": true,
2421
+ "temperature": false,
2422
+ "knowledge": "2026-01-31",
2423
+ "release_date": "2026-06-09",
2424
+ "last_updated": "2026-06-09",
2425
+ "modalities": {
2426
+ "input": [
2427
+ "text",
2428
+ "image",
2429
+ "pdf"
2430
+ ],
2431
+ "output": [
2432
+ "text"
2433
+ ]
2434
+ },
2435
+ "open_weights": false,
2436
+ "limit": {
2437
+ "context": 1000000,
2438
+ "output": 128000
2439
+ },
2440
+ "cost": {
2441
+ "input": 10,
2442
+ "output": 50,
2443
+ "cache_read": 1,
2444
+ "cache_write": 12.5
2445
+ }
2446
+ },
2447
+ "amazon.nova-lite-v1:0": {
2448
+ "id": "amazon.nova-lite-v1:0",
2449
+ "name": "Nova Lite",
2450
+ "description": "Efficient model for low-latency assistance, extraction, and routine automation",
2451
+ "family": "nova-lite",
2452
+ "attachment": true,
2453
+ "reasoning": false,
2454
+ "tool_call": true,
2455
+ "temperature": true,
2456
+ "knowledge": "2024-10",
2457
+ "release_date": "2024-12-03",
2458
+ "last_updated": "2024-12-03",
2459
+ "modalities": {
2460
+ "input": [
2461
+ "text",
2462
+ "image",
2463
+ "video"
2464
+ ],
2465
+ "output": [
2466
+ "text"
2467
+ ]
2468
+ },
2469
+ "open_weights": false,
2470
+ "limit": {
2471
+ "context": 300000,
1305
2472
  "output": 8192
1306
2473
  },
1307
2474
  "cost": {
1308
- "input": 0.6,
1309
- "output": 6
2475
+ "input": 0.06,
2476
+ "output": 0.24,
2477
+ "cache_read": 0.015
1310
2478
  }
1311
2479
  },
1312
- "minimax.minimax-m2.1": {
1313
- "id": "minimax.minimax-m2.1",
1314
- "name": "MiniMax M2.1",
1315
- "family": "minimax",
1316
- "attachment": false,
2480
+ "us.anthropic.claude-haiku-4-5-20251001-v1:0": {
2481
+ "id": "us.anthropic.claude-haiku-4-5-20251001-v1:0",
2482
+ "name": "Claude Haiku 4.5 (US)",
2483
+ "description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
2484
+ "family": "claude-haiku",
2485
+ "attachment": true,
1317
2486
  "reasoning": true,
2487
+ "reasoning_options": [
2488
+ {
2489
+ "type": "budget_tokens",
2490
+ "min": 1024
2491
+ }
2492
+ ],
1318
2493
  "tool_call": true,
1319
- "structured_output": false,
2494
+ "structured_output": true,
1320
2495
  "temperature": true,
1321
- "release_date": "2025-12-23",
1322
- "last_updated": "2025-12-23",
2496
+ "knowledge": "2025-02-28",
2497
+ "release_date": "2025-10-15",
2498
+ "last_updated": "2025-10-15",
1323
2499
  "modalities": {
1324
2500
  "input": [
2501
+ "text",
2502
+ "image",
2503
+ "pdf"
2504
+ ],
2505
+ "output": [
1325
2506
  "text"
2507
+ ]
2508
+ },
2509
+ "open_weights": false,
2510
+ "limit": {
2511
+ "context": 200000,
2512
+ "output": 64000
2513
+ },
2514
+ "cost": {
2515
+ "input": 1,
2516
+ "output": 5,
2517
+ "cache_read": 0.1,
2518
+ "cache_write": 1.25
2519
+ }
2520
+ },
2521
+ "xai.grok-4.3": {
2522
+ "id": "xai.grok-4.3",
2523
+ "name": "Grok 4.3",
2524
+ "description": "xAI's default Grok for chat, coding, agentic tools, and lower hallucination risk",
2525
+ "family": "grok",
2526
+ "attachment": true,
2527
+ "reasoning": true,
2528
+ "reasoning_options": [
2529
+ {
2530
+ "type": "effort",
2531
+ "values": [
2532
+ "none",
2533
+ "low",
2534
+ "medium",
2535
+ "high"
2536
+ ]
2537
+ }
2538
+ ],
2539
+ "tool_call": true,
2540
+ "structured_output": true,
2541
+ "temperature": true,
2542
+ "release_date": "2026-04-17",
2543
+ "last_updated": "2026-06-28",
2544
+ "modalities": {
2545
+ "input": [
2546
+ "text",
2547
+ "image"
1326
2548
  ],
1327
2549
  "output": [
1328
2550
  "text"
1329
2551
  ]
1330
2552
  },
1331
- "open_weights": true,
2553
+ "open_weights": false,
1332
2554
  "limit": {
1333
- "context": 204800,
2555
+ "context": 1000000,
1334
2556
  "output": 131072
1335
2557
  },
2558
+ "provider": {
2559
+ "npm": "@ai-sdk/amazon-bedrock/mantle",
2560
+ "api": "https://bedrock-mantle.${AWS_REGION}.api.aws/openai/v1",
2561
+ "shape": "responses"
2562
+ },
1336
2563
  "cost": {
1337
- "input": 0.3,
1338
- "output": 1.2
2564
+ "input": 1.25,
2565
+ "output": 2.5,
2566
+ "cache_read": 0.2
1339
2567
  }
1340
2568
  },
1341
- "qwen.qwen3-next-80b-a3b": {
1342
- "id": "qwen.qwen3-next-80b-a3b",
1343
- "name": "Qwen/Qwen3-Next-80B-A3B-Instruct",
1344
- "family": "qwen",
2569
+ "mistral.voxtral-mini-3b-2507": {
2570
+ "id": "mistral.voxtral-mini-3b-2507",
2571
+ "name": "Voxtral Mini 3B 2507",
2572
+ "description": "Efficient Mistral model for fast chat, extraction, and production assistants",
2573
+ "family": "mistral",
1345
2574
  "attachment": false,
1346
2575
  "reasoning": false,
1347
2576
  "tool_call": true,
1348
2577
  "structured_output": true,
1349
2578
  "temperature": true,
1350
- "release_date": "2025-09-18",
1351
- "last_updated": "2025-11-25",
2579
+ "release_date": "2024-12-01",
2580
+ "last_updated": "2024-12-01",
1352
2581
  "modalities": {
1353
2582
  "input": [
2583
+ "audio",
1354
2584
  "text"
1355
2585
  ],
1356
2586
  "output": [
@@ -1359,51 +2589,50 @@
1359
2589
  },
1360
2590
  "open_weights": false,
1361
2591
  "limit": {
1362
- "context": 262000,
1363
- "output": 262000
2592
+ "context": 128000,
2593
+ "output": 4096
1364
2594
  },
1365
2595
  "cost": {
1366
- "input": 0.14,
1367
- "output": 1.4
2596
+ "input": 0.04,
2597
+ "output": 0.04
1368
2598
  }
1369
2599
  },
1370
- "au.anthropic.claude-opus-4-6-v1": {
1371
- "id": "au.anthropic.claude-opus-4-6-v1",
1372
- "name": "AU Anthropic Claude Opus 4.6",
1373
- "family": "claude-opus",
1374
- "attachment": true,
2600
+ "moonshot.kimi-k2-thinking": {
2601
+ "id": "moonshot.kimi-k2-thinking",
2602
+ "name": "Kimi K2 Thinking",
2603
+ "description": "Kimi reasoning model for long-horizon research, planning, and tool use",
2604
+ "family": "kimi-thinking",
2605
+ "attachment": false,
1375
2606
  "reasoning": true,
2607
+ "reasoning_options": [],
1376
2608
  "tool_call": true,
2609
+ "interleaved": true,
1377
2610
  "structured_output": true,
1378
2611
  "temperature": true,
1379
- "knowledge": "2025-05",
1380
- "release_date": "2026-02-05",
1381
- "last_updated": "2026-02-05",
2612
+ "release_date": "2025-12-02",
2613
+ "last_updated": "2025-12-02",
1382
2614
  "modalities": {
1383
2615
  "input": [
1384
- "text",
1385
- "image",
1386
- "pdf"
2616
+ "text"
1387
2617
  ],
1388
2618
  "output": [
1389
2619
  "text"
1390
2620
  ]
1391
2621
  },
1392
- "open_weights": false,
2622
+ "open_weights": true,
1393
2623
  "limit": {
1394
- "context": 1000000,
1395
- "output": 128000
2624
+ "context": 262143,
2625
+ "output": 16000
1396
2626
  },
1397
2627
  "cost": {
1398
- "input": 16.5,
1399
- "output": 82.5,
1400
- "cache_read": 1.65,
1401
- "cache_write": 20.625
2628
+ "input": 0.6,
2629
+ "output": 2.5
1402
2630
  }
1403
2631
  },
1404
- "meta.llama3-1-8b-instruct-v1:0": {
1405
- "id": "meta.llama3-1-8b-instruct-v1:0",
1406
- "name": "Llama 3.1 8B Instruct",
2632
+ "meta.llama3-1-70b-instruct-v1:0": {
2633
+ "id": "meta.llama3-1-70b-instruct-v1:0",
2634
+ "name": "Llama 3.1 70B Instruct",
2635
+ "description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
1407
2636
  "family": "llama",
1408
2637
  "attachment": false,
1409
2638
  "reasoning": false,
@@ -1426,114 +2655,150 @@
1426
2655
  "output": 4096
1427
2656
  },
1428
2657
  "cost": {
1429
- "input": 0.22,
1430
- "output": 0.22
2658
+ "input": 0.72,
2659
+ "output": 0.72
1431
2660
  }
1432
2661
  },
1433
- "amazon.nova-2-lite-v1:0": {
1434
- "id": "amazon.nova-2-lite-v1:0",
1435
- "name": "Nova 2 Lite",
1436
- "family": "nova",
2662
+ "us.deepseek.r1-v1:0": {
2663
+ "id": "us.deepseek.r1-v1:0",
2664
+ "name": "DeepSeek-R1 (US)",
2665
+ "description": "Classic open reasoning model for transparent math, coding, and deliberate problem solving",
2666
+ "family": "deepseek-thinking",
1437
2667
  "attachment": false,
1438
- "reasoning": false,
2668
+ "reasoning": true,
2669
+ "reasoning_options": [],
1439
2670
  "tool_call": true,
1440
2671
  "temperature": true,
1441
- "release_date": "2024-12-01",
1442
- "last_updated": "2024-12-01",
2672
+ "knowledge": "2024-07",
2673
+ "release_date": "2025-01-20",
2674
+ "last_updated": "2025-05-29",
1443
2675
  "modalities": {
1444
2676
  "input": [
1445
- "text",
1446
- "image",
1447
- "video"
2677
+ "text"
1448
2678
  ],
1449
2679
  "output": [
1450
2680
  "text"
1451
2681
  ]
1452
2682
  },
1453
- "open_weights": false,
2683
+ "open_weights": true,
1454
2684
  "limit": {
1455
2685
  "context": 128000,
1456
- "output": 4096
2686
+ "output": 32768
1457
2687
  },
1458
2688
  "cost": {
1459
- "input": 0.33,
1460
- "output": 2.75
2689
+ "input": 1.35,
2690
+ "output": 5.4
1461
2691
  }
1462
2692
  },
1463
- "moonshotai.kimi-k2.5": {
1464
- "id": "moonshotai.kimi-k2.5",
1465
- "name": "Kimi K2.5",
1466
- "family": "kimi",
1467
- "attachment": false,
2693
+ "global.anthropic.claude-sonnet-4-6": {
2694
+ "id": "global.anthropic.claude-sonnet-4-6",
2695
+ "name": "Claude Sonnet 4.6 (Global)",
2696
+ "description": "Claude workhorse for coding agents, careful analysis, and production cost control",
2697
+ "family": "claude-sonnet",
2698
+ "attachment": true,
1468
2699
  "reasoning": true,
2700
+ "reasoning_options": [
2701
+ {
2702
+ "type": "effort",
2703
+ "values": [
2704
+ "low",
2705
+ "medium",
2706
+ "high",
2707
+ "max"
2708
+ ]
2709
+ },
2710
+ {
2711
+ "type": "budget_tokens",
2712
+ "min": 1024
2713
+ }
2714
+ ],
1469
2715
  "tool_call": true,
1470
- "interleaved": true,
1471
2716
  "structured_output": true,
1472
2717
  "temperature": true,
1473
- "release_date": "2026-02-06",
1474
- "last_updated": "2026-02-06",
2718
+ "knowledge": "2025-08-31",
2719
+ "release_date": "2026-02-17",
2720
+ "last_updated": "2026-03-13",
1475
2721
  "modalities": {
1476
2722
  "input": [
1477
2723
  "text",
1478
- "image"
2724
+ "image",
2725
+ "pdf"
1479
2726
  ],
1480
2727
  "output": [
1481
2728
  "text"
1482
2729
  ]
1483
2730
  },
1484
- "open_weights": true,
2731
+ "open_weights": false,
1485
2732
  "limit": {
1486
- "context": 256000,
1487
- "output": 256000
2733
+ "context": 1000000,
2734
+ "output": 64000
1488
2735
  },
1489
2736
  "cost": {
1490
- "input": 0.6,
1491
- "output": 3
2737
+ "input": 3,
2738
+ "output": 15,
2739
+ "cache_read": 0.3,
2740
+ "cache_write": 3.75
1492
2741
  }
1493
2742
  },
1494
- "nvidia.nemotron-super-3-120b": {
1495
- "id": "nvidia.nemotron-super-3-120b",
1496
- "name": "NVIDIA Nemotron 3 Super 120B A12B",
1497
- "family": "nemotron",
1498
- "attachment": false,
2743
+ "anthropic.claude-haiku-4-5-20251001-v1:0": {
2744
+ "id": "anthropic.claude-haiku-4-5-20251001-v1:0",
2745
+ "name": "Claude Haiku 4.5",
2746
+ "description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
2747
+ "family": "claude-haiku",
2748
+ "attachment": true,
1499
2749
  "reasoning": true,
2750
+ "reasoning_options": [
2751
+ {
2752
+ "type": "budget_tokens",
2753
+ "min": 1024
2754
+ }
2755
+ ],
1500
2756
  "tool_call": true,
1501
2757
  "structured_output": true,
1502
2758
  "temperature": true,
1503
- "release_date": "2026-03-11",
1504
- "last_updated": "2026-03-11",
2759
+ "knowledge": "2025-02-28",
2760
+ "release_date": "2025-10-15",
2761
+ "last_updated": "2025-10-15",
1505
2762
  "modalities": {
1506
2763
  "input": [
1507
- "text"
2764
+ "text",
2765
+ "image",
2766
+ "pdf"
1508
2767
  ],
1509
2768
  "output": [
1510
2769
  "text"
1511
2770
  ]
1512
2771
  },
1513
- "open_weights": true,
2772
+ "open_weights": false,
1514
2773
  "limit": {
1515
- "context": 262144,
1516
- "output": 131072
2774
+ "context": 200000,
2775
+ "output": 64000
1517
2776
  },
1518
2777
  "cost": {
1519
- "input": 0.15,
1520
- "output": 0.65
2778
+ "input": 1,
2779
+ "output": 5,
2780
+ "cache_read": 0.1,
2781
+ "cache_write": 1.25
1521
2782
  }
1522
2783
  },
1523
- "qwen.qwen3-coder-next": {
1524
- "id": "qwen.qwen3-coder-next",
1525
- "name": "Qwen3 Coder Next",
1526
- "family": "qwen",
2784
+ "moonshotai.kimi-k2.5": {
2785
+ "id": "moonshotai.kimi-k2.5",
2786
+ "name": "Kimi K2.5",
2787
+ "description": "Kimi multimodal agent model for visual understanding, coding, and planning",
2788
+ "family": "kimi",
1527
2789
  "attachment": false,
1528
2790
  "reasoning": true,
2791
+ "reasoning_options": [],
1529
2792
  "tool_call": true,
2793
+ "interleaved": true,
1530
2794
  "structured_output": true,
1531
2795
  "temperature": true,
1532
2796
  "release_date": "2026-02-06",
1533
2797
  "last_updated": "2026-02-06",
1534
2798
  "modalities": {
1535
2799
  "input": [
1536
- "text"
2800
+ "text",
2801
+ "image"
1537
2802
  ],
1538
2803
  "output": [
1539
2804
  "text"
@@ -1541,61 +2806,75 @@
1541
2806
  },
1542
2807
  "open_weights": true,
1543
2808
  "limit": {
1544
- "context": 131072,
1545
- "output": 65536
2809
+ "context": 262143,
2810
+ "output": 16000
1546
2811
  },
1547
2812
  "cost": {
1548
- "input": 0.22,
1549
- "output": 1.8
2813
+ "input": 0.6,
2814
+ "output": 3
1550
2815
  }
1551
2816
  },
1552
- "mistral.ministral-3-3b-instruct": {
1553
- "id": "mistral.ministral-3-3b-instruct",
1554
- "name": "Ministral 3 3B",
1555
- "family": "ministral",
1556
- "attachment": false,
1557
- "reasoning": false,
2817
+ "au.anthropic.claude-opus-4-8": {
2818
+ "id": "au.anthropic.claude-opus-4-8",
2819
+ "name": "Claude Opus 4.8 (AU)",
2820
+ "description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
2821
+ "family": "claude-opus",
2822
+ "attachment": true,
2823
+ "reasoning": true,
2824
+ "reasoning_options": [
2825
+ {
2826
+ "type": "effort",
2827
+ "values": [
2828
+ "low",
2829
+ "medium",
2830
+ "high",
2831
+ "xhigh",
2832
+ "max"
2833
+ ]
2834
+ }
2835
+ ],
1558
2836
  "tool_call": true,
1559
- "structured_output": true,
1560
- "temperature": true,
1561
- "release_date": "2025-12-02",
1562
- "last_updated": "2025-12-02",
2837
+ "temperature": false,
2838
+ "release_date": "2026-05-28",
2839
+ "last_updated": "2026-05-28",
1563
2840
  "modalities": {
1564
2841
  "input": [
1565
2842
  "text",
1566
- "image"
2843
+ "image",
2844
+ "pdf"
1567
2845
  ],
1568
2846
  "output": [
1569
2847
  "text"
1570
2848
  ]
1571
2849
  },
1572
- "open_weights": true,
2850
+ "open_weights": false,
1573
2851
  "limit": {
1574
- "context": 256000,
1575
- "output": 8192
2852
+ "context": 1000000,
2853
+ "output": 128000
1576
2854
  },
1577
2855
  "cost": {
1578
- "input": 0.1,
1579
- "output": 0.1
2856
+ "input": 5,
2857
+ "output": 25,
2858
+ "cache_read": 0.5,
2859
+ "cache_write": 6.25
1580
2860
  }
1581
2861
  },
1582
- "eu.anthropic.claude-haiku-4-5-20251001-v1:0": {
1583
- "id": "eu.anthropic.claude-haiku-4-5-20251001-v1:0",
1584
- "name": "Claude Haiku 4.5 (EU)",
1585
- "family": "claude-haiku",
1586
- "attachment": true,
1587
- "reasoning": true,
2862
+ "nvidia.nemotron-nano-12b-v2": {
2863
+ "id": "nvidia.nemotron-nano-12b-v2",
2864
+ "name": "NVIDIA Nemotron Nano 12B v2 VL BF16",
2865
+ "description": "Nemotron multimodal model for visual reasoning and agentic AI workflows",
2866
+ "family": "nemotron",
2867
+ "attachment": false,
2868
+ "reasoning": false,
1588
2869
  "tool_call": true,
1589
2870
  "structured_output": true,
1590
2871
  "temperature": true,
1591
- "knowledge": "2025-02-28",
1592
- "release_date": "2025-10-15",
1593
- "last_updated": "2025-10-15",
2872
+ "release_date": "2024-12-01",
2873
+ "last_updated": "2024-12-01",
1594
2874
  "modalities": {
1595
2875
  "input": [
1596
2876
  "text",
1597
- "image",
1598
- "pdf"
2877
+ "image"
1599
2878
  ],
1600
2879
  "output": [
1601
2880
  "text"
@@ -1603,27 +2882,28 @@
1603
2882
  },
1604
2883
  "open_weights": false,
1605
2884
  "limit": {
1606
- "context": 200000,
1607
- "output": 64000
2885
+ "context": 128000,
2886
+ "output": 4096
1608
2887
  },
1609
2888
  "cost": {
1610
- "input": 1,
1611
- "output": 5,
1612
- "cache_read": 0.1,
1613
- "cache_write": 1.25
2889
+ "input": 0.2,
2890
+ "output": 0.6
1614
2891
  }
1615
2892
  },
1616
- "minimax.minimax-m2": {
1617
- "id": "minimax.minimax-m2",
1618
- "name": "MiniMax M2",
1619
- "family": "minimax",
2893
+ "zai.glm-4.7-flash": {
2894
+ "id": "zai.glm-4.7-flash",
2895
+ "name": "GLM-4.7-Flash",
2896
+ "description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
2897
+ "family": "glm-flash",
1620
2898
  "attachment": false,
1621
2899
  "reasoning": true,
2900
+ "reasoning_options": [],
1622
2901
  "tool_call": true,
1623
- "structured_output": false,
2902
+ "structured_output": true,
1624
2903
  "temperature": true,
1625
- "release_date": "2025-10-27",
1626
- "last_updated": "2025-10-27",
2904
+ "knowledge": "2025-04",
2905
+ "release_date": "2026-01-19",
2906
+ "last_updated": "2026-01-19",
1627
2907
  "modalities": {
1628
2908
  "input": [
1629
2909
  "text"
@@ -1634,54 +2914,108 @@
1634
2914
  },
1635
2915
  "open_weights": true,
1636
2916
  "limit": {
1637
- "context": 204608,
1638
- "output": 128000
2917
+ "context": 200000,
2918
+ "output": 131072
1639
2919
  },
1640
2920
  "cost": {
1641
- "input": 0.3,
1642
- "output": 1.2
2921
+ "input": 0.07,
2922
+ "output": 0.4
1643
2923
  }
1644
2924
  },
1645
- "meta.llama3-1-70b-instruct-v1:0": {
1646
- "id": "meta.llama3-1-70b-instruct-v1:0",
1647
- "name": "Llama 3.1 70B Instruct",
2925
+ "meta.llama4-scout-17b-instruct-v1:0": {
2926
+ "id": "meta.llama4-scout-17b-instruct-v1:0",
2927
+ "name": "Llama 4 Scout 17B Instruct",
2928
+ "description": "Open multimodal Llama model for long-context analysis and efficient agents",
1648
2929
  "family": "llama",
1649
- "attachment": false,
2930
+ "attachment": true,
1650
2931
  "reasoning": false,
1651
2932
  "tool_call": true,
1652
- "temperature": true,
1653
- "knowledge": "2023-12",
1654
- "release_date": "2024-07-23",
1655
- "last_updated": "2024-07-23",
2933
+ "temperature": true,
2934
+ "knowledge": "2024-08",
2935
+ "release_date": "2025-04-05",
2936
+ "last_updated": "2025-04-05",
2937
+ "modalities": {
2938
+ "input": [
2939
+ "text",
2940
+ "image"
2941
+ ],
2942
+ "output": [
2943
+ "text"
2944
+ ]
2945
+ },
2946
+ "open_weights": true,
2947
+ "limit": {
2948
+ "context": 3500000,
2949
+ "output": 16384
2950
+ },
2951
+ "cost": {
2952
+ "input": 0.17,
2953
+ "output": 0.66
2954
+ }
2955
+ },
2956
+ "au.anthropic.claude-sonnet-5": {
2957
+ "id": "au.anthropic.claude-sonnet-5",
2958
+ "name": "Claude Sonnet 5 (AU)",
2959
+ "description": "Everyday Claude agent model for coding, planning, browsing, and general work",
2960
+ "family": "claude-sonnet",
2961
+ "attachment": true,
2962
+ "reasoning": true,
2963
+ "reasoning_options": [
2964
+ {
2965
+ "type": "toggle"
2966
+ },
2967
+ {
2968
+ "type": "effort",
2969
+ "values": [
2970
+ "low",
2971
+ "medium",
2972
+ "high",
2973
+ "xhigh",
2974
+ "max"
2975
+ ]
2976
+ }
2977
+ ],
2978
+ "tool_call": true,
2979
+ "structured_output": true,
2980
+ "temperature": false,
2981
+ "knowledge": "2026-01-31",
2982
+ "release_date": "2026-06-30",
2983
+ "last_updated": "2026-06-30",
1656
2984
  "modalities": {
1657
2985
  "input": [
1658
- "text"
2986
+ "text",
2987
+ "image",
2988
+ "pdf"
1659
2989
  ],
1660
2990
  "output": [
1661
2991
  "text"
1662
2992
  ]
1663
2993
  },
1664
- "open_weights": true,
2994
+ "open_weights": false,
1665
2995
  "limit": {
1666
- "context": 128000,
1667
- "output": 4096
2996
+ "context": 1000000,
2997
+ "output": 128000
1668
2998
  },
1669
2999
  "cost": {
1670
- "input": 0.72,
1671
- "output": 0.72
3000
+ "input": 2,
3001
+ "output": 10,
3002
+ "cache_read": 0.2,
3003
+ "cache_write": 2.5
1672
3004
  }
1673
3005
  },
1674
- "openai.gpt-oss-safeguard-120b": {
1675
- "id": "openai.gpt-oss-safeguard-120b",
1676
- "name": "GPT OSS Safeguard 120B",
1677
- "family": "gpt-oss",
3006
+ "qwen.qwen3-235b-a22b-2507-v1:0": {
3007
+ "id": "qwen.qwen3-235b-a22b-2507-v1:0",
3008
+ "name": "Qwen3 235B A22B 2507",
3009
+ "description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
3010
+ "family": "qwen",
1678
3011
  "attachment": false,
1679
3012
  "reasoning": false,
1680
3013
  "tool_call": true,
1681
3014
  "structured_output": true,
1682
3015
  "temperature": true,
1683
- "release_date": "2025-10-29",
1684
- "last_updated": "2025-10-29",
3016
+ "knowledge": "2024-04",
3017
+ "release_date": "2025-09-18",
3018
+ "last_updated": "2025-09-18",
1685
3019
  "modalities": {
1686
3020
  "input": [
1687
3021
  "text"
@@ -1690,22 +3024,29 @@
1690
3024
  "text"
1691
3025
  ]
1692
3026
  },
1693
- "open_weights": false,
3027
+ "open_weights": true,
1694
3028
  "limit": {
1695
- "context": 128000,
1696
- "output": 16384
3029
+ "context": 262144,
3030
+ "output": 131072
1697
3031
  },
1698
3032
  "cost": {
1699
- "input": 0.15,
1700
- "output": 0.6
3033
+ "input": 0.22,
3034
+ "output": 0.88
1701
3035
  }
1702
3036
  },
1703
- "global.anthropic.claude-haiku-4-5-20251001-v1:0": {
1704
- "id": "global.anthropic.claude-haiku-4-5-20251001-v1:0",
1705
- "name": "Claude Haiku 4.5 (Global)",
3037
+ "eu.anthropic.claude-haiku-4-5-20251001-v1:0": {
3038
+ "id": "eu.anthropic.claude-haiku-4-5-20251001-v1:0",
3039
+ "name": "Claude Haiku 4.5 (EU)",
3040
+ "description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
1706
3041
  "family": "claude-haiku",
1707
3042
  "attachment": true,
1708
3043
  "reasoning": true,
3044
+ "reasoning_options": [
3045
+ {
3046
+ "type": "budget_tokens",
3047
+ "min": 1024
3048
+ }
3049
+ ],
1709
3050
  "tool_call": true,
1710
3051
  "structured_output": true,
1711
3052
  "temperature": true,
@@ -1728,23 +3069,34 @@
1728
3069
  "output": 64000
1729
3070
  },
1730
3071
  "cost": {
1731
- "input": 1,
1732
- "output": 5,
1733
- "cache_read": 0.1,
1734
- "cache_write": 1.25
3072
+ "input": 1.1,
3073
+ "output": 5.5,
3074
+ "cache_read": 0.11,
3075
+ "cache_write": 1.375
1735
3076
  }
1736
3077
  },
1737
- "amazon.nova-micro-v1:0": {
1738
- "id": "amazon.nova-micro-v1:0",
1739
- "name": "Nova Micro",
1740
- "family": "nova-micro",
3078
+ "openai.gpt-oss-20b-1:0": {
3079
+ "id": "openai.gpt-oss-20b-1:0",
3080
+ "name": "gpt-oss-20b",
3081
+ "description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
3082
+ "family": "gpt-oss",
1741
3083
  "attachment": false,
1742
- "reasoning": false,
3084
+ "reasoning": true,
3085
+ "reasoning_options": [
3086
+ {
3087
+ "type": "effort",
3088
+ "values": [
3089
+ "low",
3090
+ "medium",
3091
+ "high"
3092
+ ]
3093
+ }
3094
+ ],
1743
3095
  "tool_call": true,
3096
+ "structured_output": true,
1744
3097
  "temperature": true,
1745
- "knowledge": "2024-10",
1746
- "release_date": "2024-12-03",
1747
- "last_updated": "2024-12-03",
3098
+ "release_date": "2025-08-05",
3099
+ "last_updated": "2025-08-05",
1748
3100
  "modalities": {
1749
3101
  "input": [
1750
3102
  "text"
@@ -1756,26 +3108,36 @@
1756
3108
  "open_weights": false,
1757
3109
  "limit": {
1758
3110
  "context": 128000,
1759
- "output": 8192
3111
+ "output": 16384
1760
3112
  },
1761
3113
  "cost": {
1762
- "input": 0.035,
1763
- "output": 0.14,
1764
- "cache_read": 0.00875
3114
+ "input": 0.07,
3115
+ "output": 0.3
1765
3116
  }
1766
3117
  },
1767
- "anthropic.claude-sonnet-4-5-20250929-v1:0": {
1768
- "id": "anthropic.claude-sonnet-4-5-20250929-v1:0",
1769
- "name": "Claude Sonnet 4.5",
1770
- "family": "claude-sonnet",
3118
+ "jp.anthropic.claude-opus-4-8": {
3119
+ "id": "jp.anthropic.claude-opus-4-8",
3120
+ "name": "Claude Opus 4.8 (JP)",
3121
+ "description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
3122
+ "family": "claude-opus",
1771
3123
  "attachment": true,
1772
3124
  "reasoning": true,
3125
+ "reasoning_options": [
3126
+ {
3127
+ "type": "effort",
3128
+ "values": [
3129
+ "low",
3130
+ "medium",
3131
+ "high",
3132
+ "xhigh",
3133
+ "max"
3134
+ ]
3135
+ }
3136
+ ],
1773
3137
  "tool_call": true,
1774
- "structured_output": true,
1775
- "temperature": true,
1776
- "knowledge": "2025-07-31",
1777
- "release_date": "2025-09-29",
1778
- "last_updated": "2025-09-29",
3138
+ "temperature": false,
3139
+ "release_date": "2026-05-28",
3140
+ "last_updated": "2026-05-28",
1779
3141
  "modalities": {
1780
3142
  "input": [
1781
3143
  "text",
@@ -1788,28 +3150,39 @@
1788
3150
  },
1789
3151
  "open_weights": false,
1790
3152
  "limit": {
1791
- "context": 200000,
1792
- "output": 64000
3153
+ "context": 1000000,
3154
+ "output": 128000
1793
3155
  },
1794
3156
  "cost": {
1795
- "input": 3,
1796
- "output": 15,
1797
- "cache_read": 0.3,
1798
- "cache_write": 3.75
3157
+ "input": 5,
3158
+ "output": 25,
3159
+ "cache_read": 0.5,
3160
+ "cache_write": 6.25
1799
3161
  }
1800
3162
  },
1801
- "eu.anthropic.claude-sonnet-4-5-20250929-v1:0": {
1802
- "id": "eu.anthropic.claude-sonnet-4-5-20250929-v1:0",
1803
- "name": "Claude Sonnet 4.5 (EU)",
1804
- "family": "claude-sonnet",
3163
+ "anthropic.claude-opus-4-8": {
3164
+ "id": "anthropic.claude-opus-4-8",
3165
+ "name": "Claude Opus 4.8",
3166
+ "description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
3167
+ "family": "claude-opus",
1805
3168
  "attachment": true,
1806
3169
  "reasoning": true,
3170
+ "reasoning_options": [
3171
+ {
3172
+ "type": "effort",
3173
+ "values": [
3174
+ "low",
3175
+ "medium",
3176
+ "high",
3177
+ "xhigh",
3178
+ "max"
3179
+ ]
3180
+ }
3181
+ ],
1807
3182
  "tool_call": true,
1808
- "structured_output": true,
1809
- "temperature": true,
1810
- "knowledge": "2025-07-31",
1811
- "release_date": "2025-09-29",
1812
- "last_updated": "2025-09-29",
3183
+ "temperature": false,
3184
+ "release_date": "2026-05-28",
3185
+ "last_updated": "2026-05-28",
1813
3186
  "modalities": {
1814
3187
  "input": [
1815
3188
  "text",
@@ -1822,57 +3195,59 @@
1822
3195
  },
1823
3196
  "open_weights": false,
1824
3197
  "limit": {
1825
- "context": 200000,
1826
- "output": 64000
3198
+ "context": 1000000,
3199
+ "output": 128000
1827
3200
  },
1828
3201
  "cost": {
1829
- "input": 3,
1830
- "output": 15,
1831
- "cache_read": 0.3,
1832
- "cache_write": 3.75
3202
+ "input": 5,
3203
+ "output": 25,
3204
+ "cache_read": 0.5,
3205
+ "cache_write": 6.25
1833
3206
  }
1834
3207
  },
1835
- "mistral.voxtral-small-24b-2507": {
1836
- "id": "mistral.voxtral-small-24b-2507",
1837
- "name": "Voxtral Small 24B 2507",
1838
- "family": "mistral",
1839
- "attachment": true,
3208
+ "qwen.qwen3-coder-30b-a3b-v1:0": {
3209
+ "id": "qwen.qwen3-coder-30b-a3b-v1:0",
3210
+ "name": "Qwen3 Coder 30B A3B Instruct",
3211
+ "description": "Qwen coding model for software agents, repository edits, and code reasoning",
3212
+ "family": "qwen",
3213
+ "attachment": false,
1840
3214
  "reasoning": false,
1841
3215
  "tool_call": true,
1842
3216
  "structured_output": true,
1843
3217
  "temperature": true,
1844
- "release_date": "2025-07-01",
1845
- "last_updated": "2025-07-01",
3218
+ "knowledge": "2024-04",
3219
+ "release_date": "2025-09-18",
3220
+ "last_updated": "2025-09-18",
1846
3221
  "modalities": {
1847
3222
  "input": [
1848
- "text",
1849
- "audio"
3223
+ "text"
1850
3224
  ],
1851
3225
  "output": [
1852
3226
  "text"
1853
3227
  ]
1854
3228
  },
1855
- "open_weights": true,
3229
+ "open_weights": false,
1856
3230
  "limit": {
1857
- "context": 32000,
1858
- "output": 8192
3231
+ "context": 262144,
3232
+ "output": 131072
1859
3233
  },
1860
3234
  "cost": {
1861
3235
  "input": 0.15,
1862
- "output": 0.35
3236
+ "output": 0.6
1863
3237
  }
1864
3238
  },
1865
- "nvidia.nemotron-nano-3-30b": {
1866
- "id": "nvidia.nemotron-nano-3-30b",
1867
- "name": "NVIDIA Nemotron Nano 3 30B",
1868
- "family": "nemotron",
3239
+ "qwen.qwen3-next-80b-a3b": {
3240
+ "id": "qwen.qwen3-next-80b-a3b",
3241
+ "name": "Qwen/Qwen3-Next-80B-A3B-Instruct",
3242
+ "description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
3243
+ "family": "qwen",
1869
3244
  "attachment": false,
1870
- "reasoning": true,
3245
+ "reasoning": false,
1871
3246
  "tool_call": true,
1872
3247
  "structured_output": true,
1873
3248
  "temperature": true,
1874
- "release_date": "2025-12-23",
1875
- "last_updated": "2025-12-23",
3249
+ "release_date": "2025-09-18",
3250
+ "last_updated": "2025-11-25",
1876
3251
  "modalities": {
1877
3252
  "input": [
1878
3253
  "text"
@@ -1881,28 +3256,41 @@
1881
3256
  "text"
1882
3257
  ]
1883
3258
  },
1884
- "open_weights": true,
3259
+ "open_weights": false,
1885
3260
  "limit": {
1886
- "context": 128000,
1887
- "output": 4096
3261
+ "context": 262000,
3262
+ "output": 262000
1888
3263
  },
1889
3264
  "cost": {
1890
- "input": 0.06,
1891
- "output": 0.24
3265
+ "input": 0.14,
3266
+ "output": 1.4
1892
3267
  }
1893
3268
  },
1894
- "anthropic.claude-haiku-4-5-20251001-v1:0": {
1895
- "id": "anthropic.claude-haiku-4-5-20251001-v1:0",
1896
- "name": "Claude Haiku 4.5",
1897
- "family": "claude-haiku",
3269
+ "openai.gpt-5.5": {
3270
+ "id": "openai.gpt-5.5",
3271
+ "name": "GPT-5.5",
3272
+ "description": "Default frontier GPT for coding, computer use, research, and knowledge work",
3273
+ "family": "gpt",
1898
3274
  "attachment": true,
1899
3275
  "reasoning": true,
3276
+ "reasoning_options": [
3277
+ {
3278
+ "type": "effort",
3279
+ "values": [
3280
+ "none",
3281
+ "low",
3282
+ "medium",
3283
+ "high",
3284
+ "xhigh"
3285
+ ]
3286
+ }
3287
+ ],
1900
3288
  "tool_call": true,
1901
3289
  "structured_output": true,
1902
- "temperature": true,
1903
- "knowledge": "2025-02-28",
1904
- "release_date": "2025-10-15",
1905
- "last_updated": "2025-10-15",
3290
+ "temperature": false,
3291
+ "knowledge": "2025-12-01",
3292
+ "release_date": "2026-04-23",
3293
+ "last_updated": "2026-06-01",
1906
3294
  "modalities": {
1907
3295
  "input": [
1908
3296
  "text",
@@ -1915,30 +3303,53 @@
1915
3303
  },
1916
3304
  "open_weights": false,
1917
3305
  "limit": {
1918
- "context": 200000,
1919
- "output": 64000
3306
+ "context": 272000,
3307
+ "output": 128000
3308
+ },
3309
+ "provider": {
3310
+ "npm": "@ai-sdk/amazon-bedrock/mantle",
3311
+ "api": "https://bedrock-mantle.${AWS_REGION}.api.aws/openai/v1",
3312
+ "shape": "responses"
1920
3313
  },
1921
3314
  "cost": {
1922
- "input": 1,
1923
- "output": 5,
1924
- "cache_read": 0.1,
1925
- "cache_write": 1.25
3315
+ "input": 5.5,
3316
+ "output": 33,
3317
+ "cache_read": 0.55
1926
3318
  }
1927
3319
  },
1928
- "google.gemma-3-4b-it": {
1929
- "id": "google.gemma-3-4b-it",
1930
- "name": "Gemma 3 4B IT",
1931
- "family": "gemma",
1932
- "attachment": false,
1933
- "reasoning": false,
3320
+ "au.anthropic.claude-sonnet-4-6": {
3321
+ "id": "au.anthropic.claude-sonnet-4-6",
3322
+ "name": "AU Anthropic Claude Sonnet 4.6",
3323
+ "description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
3324
+ "family": "claude-sonnet",
3325
+ "attachment": true,
3326
+ "reasoning": true,
3327
+ "reasoning_options": [
3328
+ {
3329
+ "type": "effort",
3330
+ "values": [
3331
+ "low",
3332
+ "medium",
3333
+ "high",
3334
+ "max"
3335
+ ]
3336
+ },
3337
+ {
3338
+ "type": "budget_tokens",
3339
+ "min": 1024
3340
+ }
3341
+ ],
1934
3342
  "tool_call": true,
3343
+ "structured_output": true,
1935
3344
  "temperature": true,
1936
- "release_date": "2024-12-01",
1937
- "last_updated": "2024-12-01",
3345
+ "knowledge": "2025-08",
3346
+ "release_date": "2026-02-17",
3347
+ "last_updated": "2026-02-17",
1938
3348
  "modalities": {
1939
3349
  "input": [
1940
3350
  "text",
1941
- "image"
3351
+ "image",
3352
+ "pdf"
1942
3353
  ],
1943
3354
  "output": [
1944
3355
  "text"
@@ -1946,25 +3357,93 @@
1946
3357
  },
1947
3358
  "open_weights": false,
1948
3359
  "limit": {
1949
- "context": 128000,
1950
- "output": 4096
3360
+ "context": 1000000,
3361
+ "output": 128000
1951
3362
  },
1952
3363
  "cost": {
1953
- "input": 0.04,
1954
- "output": 0.08
3364
+ "input": 3.3,
3365
+ "output": 16.5,
3366
+ "cache_read": 0.33,
3367
+ "cache_write": 4.125
1955
3368
  }
1956
3369
  },
1957
- "global.anthropic.claude-opus-4-7": {
1958
- "id": "global.anthropic.claude-opus-4-7",
1959
- "name": "Claude Opus 4.7 (Global)",
3370
+ "us.anthropic.claude-sonnet-5": {
3371
+ "id": "us.anthropic.claude-sonnet-5",
3372
+ "name": "Claude Sonnet 5 (US)",
3373
+ "description": "Everyday Claude agent model for coding, planning, browsing, and general work",
3374
+ "family": "claude-sonnet",
3375
+ "attachment": true,
3376
+ "reasoning": true,
3377
+ "reasoning_options": [
3378
+ {
3379
+ "type": "toggle"
3380
+ },
3381
+ {
3382
+ "type": "effort",
3383
+ "values": [
3384
+ "low",
3385
+ "medium",
3386
+ "high",
3387
+ "xhigh",
3388
+ "max"
3389
+ ]
3390
+ }
3391
+ ],
3392
+ "tool_call": true,
3393
+ "structured_output": true,
3394
+ "temperature": false,
3395
+ "knowledge": "2026-01-31",
3396
+ "release_date": "2026-06-30",
3397
+ "last_updated": "2026-06-30",
3398
+ "modalities": {
3399
+ "input": [
3400
+ "text",
3401
+ "image",
3402
+ "pdf"
3403
+ ],
3404
+ "output": [
3405
+ "text"
3406
+ ]
3407
+ },
3408
+ "open_weights": false,
3409
+ "limit": {
3410
+ "context": 1000000,
3411
+ "output": 128000
3412
+ },
3413
+ "cost": {
3414
+ "input": 2,
3415
+ "output": 10,
3416
+ "cache_read": 0.2,
3417
+ "cache_write": 2.5
3418
+ }
3419
+ },
3420
+ "us.anthropic.claude-opus-4-5-20251101-v1:0": {
3421
+ "id": "us.anthropic.claude-opus-4-5-20251101-v1:0",
3422
+ "name": "Claude Opus 4.5 (US)",
3423
+ "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
1960
3424
  "family": "claude-opus",
1961
3425
  "attachment": true,
1962
3426
  "reasoning": true,
3427
+ "reasoning_options": [
3428
+ {
3429
+ "type": "effort",
3430
+ "values": [
3431
+ "low",
3432
+ "medium",
3433
+ "high"
3434
+ ]
3435
+ },
3436
+ {
3437
+ "type": "budget_tokens",
3438
+ "min": 1024
3439
+ }
3440
+ ],
1963
3441
  "tool_call": true,
1964
- "temperature": false,
1965
- "knowledge": "2026-01-31",
1966
- "release_date": "2026-04-16",
1967
- "last_updated": "2026-04-16",
3442
+ "structured_output": true,
3443
+ "temperature": true,
3444
+ "knowledge": "2025-03-31",
3445
+ "release_date": "2025-11-24",
3446
+ "last_updated": "2025-08-01",
1968
3447
  "modalities": {
1969
3448
  "input": [
1970
3449
  "text",
@@ -1977,8 +3456,8 @@
1977
3456
  },
1978
3457
  "open_weights": false,
1979
3458
  "limit": {
1980
- "context": 1000000,
1981
- "output": 128000
3459
+ "context": 200000,
3460
+ "output": 64000
1982
3461
  },
1983
3462
  "cost": {
1984
3463
  "input": 5,
@@ -1990,9 +3469,11 @@
1990
3469
  "minimax.minimax-m2.5": {
1991
3470
  "id": "minimax.minimax-m2.5",
1992
3471
  "name": "MiniMax M2.5",
3472
+ "description": "MiniMax model for chat, coding, office work, and agentic tasks",
1993
3473
  "family": "minimax",
1994
3474
  "attachment": false,
1995
3475
  "reasoning": true,
3476
+ "reasoning_options": [],
1996
3477
  "tool_call": true,
1997
3478
  "structured_output": false,
1998
3479
  "temperature": true,
@@ -2016,77 +3497,75 @@
2016
3497
  "output": 1.2
2017
3498
  }
2018
3499
  },
2019
- "openai.gpt-oss-safeguard-20b": {
2020
- "id": "openai.gpt-oss-safeguard-20b",
2021
- "name": "GPT OSS Safeguard 20B",
2022
- "family": "gpt-oss",
2023
- "attachment": false,
2024
- "reasoning": false,
2025
- "tool_call": true,
2026
- "structured_output": true,
2027
- "temperature": true,
2028
- "release_date": "2025-10-29",
2029
- "last_updated": "2025-10-29",
2030
- "modalities": {
2031
- "input": [
2032
- "text"
2033
- ],
2034
- "output": [
2035
- "text"
2036
- ]
2037
- },
2038
- "open_weights": false,
2039
- "limit": {
2040
- "context": 128000,
2041
- "output": 16384
2042
- },
2043
- "cost": {
2044
- "input": 0.07,
2045
- "output": 0.2
2046
- }
2047
- },
2048
- "deepseek.v3.2": {
2049
- "id": "deepseek.v3.2",
2050
- "name": "DeepSeek-V3.2",
2051
- "family": "deepseek",
2052
- "attachment": false,
3500
+ "eu.anthropic.claude-opus-4-8": {
3501
+ "id": "eu.anthropic.claude-opus-4-8",
3502
+ "name": "Claude Opus 4.8 (EU)",
3503
+ "description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
3504
+ "family": "claude-opus",
3505
+ "attachment": true,
2053
3506
  "reasoning": true,
3507
+ "reasoning_options": [
3508
+ {
3509
+ "type": "effort",
3510
+ "values": [
3511
+ "low",
3512
+ "medium",
3513
+ "high",
3514
+ "xhigh",
3515
+ "max"
3516
+ ]
3517
+ }
3518
+ ],
2054
3519
  "tool_call": true,
2055
- "structured_output": true,
2056
- "temperature": true,
2057
- "knowledge": "2024-07",
2058
- "release_date": "2026-02-06",
2059
- "last_updated": "2026-02-06",
3520
+ "temperature": false,
3521
+ "release_date": "2026-05-28",
3522
+ "last_updated": "2026-05-28",
2060
3523
  "modalities": {
2061
3524
  "input": [
2062
- "text"
3525
+ "text",
3526
+ "image",
3527
+ "pdf"
2063
3528
  ],
2064
3529
  "output": [
2065
3530
  "text"
2066
3531
  ]
2067
3532
  },
2068
- "open_weights": true,
3533
+ "open_weights": false,
2069
3534
  "limit": {
2070
- "context": 163840,
2071
- "output": 81920
3535
+ "context": 1000000,
3536
+ "output": 128000
2072
3537
  },
2073
3538
  "cost": {
2074
- "input": 0.62,
2075
- "output": 1.85
3539
+ "input": 5.5,
3540
+ "output": 27.5,
3541
+ "cache_read": 0.55,
3542
+ "cache_write": 6.875
2076
3543
  }
2077
3544
  },
2078
- "global.anthropic.claude-opus-4-6-v1": {
2079
- "id": "global.anthropic.claude-opus-4-6-v1",
2080
- "name": "Claude Opus 4.6 (Global)",
3545
+ "eu.anthropic.claude-opus-4-7": {
3546
+ "id": "eu.anthropic.claude-opus-4-7",
3547
+ "name": "Claude Opus 4.7 (EU)",
3548
+ "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
2081
3549
  "family": "claude-opus",
2082
3550
  "attachment": true,
2083
3551
  "reasoning": true,
3552
+ "reasoning_options": [
3553
+ {
3554
+ "type": "effort",
3555
+ "values": [
3556
+ "low",
3557
+ "medium",
3558
+ "high",
3559
+ "xhigh",
3560
+ "max"
3561
+ ]
3562
+ }
3563
+ ],
2084
3564
  "tool_call": true,
2085
- "structured_output": true,
2086
- "temperature": true,
2087
- "knowledge": "2025-05-31",
2088
- "release_date": "2026-02-05",
2089
- "last_updated": "2026-03-13",
3565
+ "temperature": false,
3566
+ "knowledge": "2026-01-31",
3567
+ "release_date": "2026-04-16",
3568
+ "last_updated": "2026-04-16",
2090
3569
  "modalities": {
2091
3570
  "input": [
2092
3571
  "text",
@@ -2103,58 +3582,60 @@
2103
3582
  "output": 128000
2104
3583
  },
2105
3584
  "cost": {
2106
- "input": 5,
2107
- "output": 25,
2108
- "cache_read": 0.5,
2109
- "cache_write": 6.25
3585
+ "input": 5.5,
3586
+ "output": 27.5,
3587
+ "cache_read": 0.55,
3588
+ "cache_write": 6.875
2110
3589
  }
2111
3590
  },
2112
- "anthropic.claude-sonnet-4-6": {
2113
- "id": "anthropic.claude-sonnet-4-6",
2114
- "name": "Claude Sonnet 4.6",
2115
- "family": "claude-sonnet",
2116
- "attachment": true,
2117
- "reasoning": true,
3591
+ "meta.llama3-1-8b-instruct-v1:0": {
3592
+ "id": "meta.llama3-1-8b-instruct-v1:0",
3593
+ "name": "Llama 3.1 8B Instruct",
3594
+ "description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
3595
+ "family": "llama",
3596
+ "attachment": false,
3597
+ "reasoning": false,
2118
3598
  "tool_call": true,
2119
- "structured_output": true,
2120
3599
  "temperature": true,
2121
- "knowledge": "2025-08-31",
2122
- "release_date": "2026-02-17",
2123
- "last_updated": "2026-03-13",
3600
+ "knowledge": "2023-12",
3601
+ "release_date": "2024-07-23",
3602
+ "last_updated": "2024-07-23",
2124
3603
  "modalities": {
2125
3604
  "input": [
2126
- "text",
2127
- "image",
2128
- "pdf"
3605
+ "text"
2129
3606
  ],
2130
3607
  "output": [
2131
3608
  "text"
2132
3609
  ]
2133
3610
  },
2134
- "open_weights": false,
3611
+ "open_weights": true,
2135
3612
  "limit": {
2136
- "context": 1000000,
2137
- "output": 64000
3613
+ "context": 128000,
3614
+ "output": 4096
2138
3615
  },
2139
3616
  "cost": {
2140
- "input": 3,
2141
- "output": 15,
2142
- "cache_read": 0.3,
2143
- "cache_write": 3.75
3617
+ "input": 0.22,
3618
+ "output": 0.22
2144
3619
  }
2145
3620
  },
2146
- "global.anthropic.claude-sonnet-4-6": {
2147
- "id": "global.anthropic.claude-sonnet-4-6",
2148
- "name": "Claude Sonnet 4.6 (Global)",
2149
- "family": "claude-sonnet",
3621
+ "us.anthropic.claude-opus-4-1-20250805-v1:0": {
3622
+ "id": "us.anthropic.claude-opus-4-1-20250805-v1:0",
3623
+ "name": "Claude Opus 4.1 (US)",
3624
+ "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
3625
+ "family": "claude-opus",
2150
3626
  "attachment": true,
2151
3627
  "reasoning": true,
3628
+ "reasoning_options": [
3629
+ {
3630
+ "type": "budget_tokens",
3631
+ "min": 1024
3632
+ }
3633
+ ],
2152
3634
  "tool_call": true,
2153
- "structured_output": true,
2154
3635
  "temperature": true,
2155
- "knowledge": "2025-08-31",
2156
- "release_date": "2026-02-17",
2157
- "last_updated": "2026-03-13",
3636
+ "knowledge": "2025-03-31",
3637
+ "release_date": "2025-08-05",
3638
+ "last_updated": "2025-08-05",
2158
3639
  "modalities": {
2159
3640
  "input": [
2160
3641
  "text",
@@ -2167,19 +3648,20 @@
2167
3648
  },
2168
3649
  "open_weights": false,
2169
3650
  "limit": {
2170
- "context": 1000000,
2171
- "output": 64000
3651
+ "context": 200000,
3652
+ "output": 32000
2172
3653
  },
2173
3654
  "cost": {
2174
- "input": 3,
2175
- "output": 15,
2176
- "cache_read": 0.3,
2177
- "cache_write": 3.75
3655
+ "input": 15,
3656
+ "output": 75,
3657
+ "cache_read": 1.5,
3658
+ "cache_write": 18.75
2178
3659
  }
2179
3660
  },
2180
- "us.meta.llama4-maverick-17b-instruct-v1:0": {
2181
- "id": "us.meta.llama4-maverick-17b-instruct-v1:0",
2182
- "name": "Llama 4 Maverick 17B Instruct (US)",
3661
+ "meta.llama4-maverick-17b-instruct-v1:0": {
3662
+ "id": "meta.llama4-maverick-17b-instruct-v1:0",
3663
+ "name": "Llama 4 Maverick 17B Instruct",
3664
+ "description": "Open multimodal Llama model for strong reasoning and fast responses",
2183
3665
  "family": "llama",
2184
3666
  "attachment": true,
2185
3667
  "reasoning": false,
@@ -2207,80 +3689,33 @@
2207
3689
  "output": 0.97
2208
3690
  }
2209
3691
  },
2210
- "us.deepseek.r1-v1:0": {
2211
- "id": "us.deepseek.r1-v1:0",
2212
- "name": "DeepSeek-R1 (US)",
2213
- "family": "deepseek-thinking",
2214
- "attachment": false,
2215
- "reasoning": true,
2216
- "tool_call": true,
2217
- "temperature": true,
2218
- "knowledge": "2024-07",
2219
- "release_date": "2025-01-20",
2220
- "last_updated": "2025-05-29",
2221
- "modalities": {
2222
- "input": [
2223
- "text"
2224
- ],
2225
- "output": [
2226
- "text"
2227
- ]
2228
- },
2229
- "open_weights": false,
2230
- "limit": {
2231
- "context": 128000,
2232
- "output": 32768
2233
- },
2234
- "cost": {
2235
- "input": 1.35,
2236
- "output": 5.4
2237
- }
2238
- },
2239
- "au.anthropic.claude-sonnet-4-5-20250929-v1:0": {
2240
- "id": "au.anthropic.claude-sonnet-4-5-20250929-v1:0",
2241
- "name": "Claude Sonnet 4.5 (AU)",
2242
- "family": "claude-sonnet",
2243
- "attachment": true,
2244
- "reasoning": true,
2245
- "tool_call": true,
2246
- "structured_output": true,
2247
- "temperature": true,
2248
- "knowledge": "2025-07-31",
2249
- "release_date": "2025-09-29",
2250
- "last_updated": "2025-09-29",
2251
- "modalities": {
2252
- "input": [
2253
- "text",
2254
- "image",
2255
- "pdf"
2256
- ],
2257
- "output": [
2258
- "text"
2259
- ]
2260
- },
2261
- "open_weights": false,
2262
- "limit": {
2263
- "context": 200000,
2264
- "output": 64000
2265
- },
2266
- "cost": {
2267
- "input": 3,
2268
- "output": 15,
2269
- "cache_read": 0.3,
2270
- "cache_write": 3.75
2271
- }
2272
- },
2273
- "us.anthropic.claude-opus-4-1-20250805-v1:0": {
2274
- "id": "us.anthropic.claude-opus-4-1-20250805-v1:0",
2275
- "name": "Claude Opus 4.1 (US)",
3692
+ "global.anthropic.claude-opus-4-5-20251101-v1:0": {
3693
+ "id": "global.anthropic.claude-opus-4-5-20251101-v1:0",
3694
+ "name": "Claude Opus 4.5 (Global)",
3695
+ "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
2276
3696
  "family": "claude-opus",
2277
3697
  "attachment": true,
2278
3698
  "reasoning": true,
3699
+ "reasoning_options": [
3700
+ {
3701
+ "type": "effort",
3702
+ "values": [
3703
+ "low",
3704
+ "medium",
3705
+ "high"
3706
+ ]
3707
+ },
3708
+ {
3709
+ "type": "budget_tokens",
3710
+ "min": 1024
3711
+ }
3712
+ ],
2279
3713
  "tool_call": true,
3714
+ "structured_output": true,
2280
3715
  "temperature": true,
2281
3716
  "knowledge": "2025-03-31",
2282
- "release_date": "2025-08-05",
2283
- "last_updated": "2025-08-05",
3717
+ "release_date": "2025-11-24",
3718
+ "last_updated": "2025-08-01",
2284
3719
  "modalities": {
2285
3720
  "input": [
2286
3721
  "text",
@@ -2294,60 +3729,74 @@
2294
3729
  "open_weights": false,
2295
3730
  "limit": {
2296
3731
  "context": 200000,
2297
- "output": 32000
3732
+ "output": 64000
2298
3733
  },
2299
3734
  "cost": {
2300
- "input": 15,
2301
- "output": 75,
2302
- "cache_read": 1.5,
2303
- "cache_write": 18.75
3735
+ "input": 5,
3736
+ "output": 25,
3737
+ "cache_read": 0.5,
3738
+ "cache_write": 6.25
2304
3739
  }
2305
3740
  },
2306
- "anthropic.claude-opus-4-1-20250805-v1:0": {
2307
- "id": "anthropic.claude-opus-4-1-20250805-v1:0",
2308
- "name": "Claude Opus 4.1",
2309
- "family": "claude-opus",
2310
- "attachment": true,
3741
+ "nvidia.nemotron-super-3-120b": {
3742
+ "id": "nvidia.nemotron-super-3-120b",
3743
+ "name": "NVIDIA Nemotron 3 Super 120B A12B",
3744
+ "description": "Nemotron middle tier for collaborative agents and high-volume reasoning workloads",
3745
+ "family": "nemotron",
3746
+ "attachment": false,
2311
3747
  "reasoning": true,
3748
+ "reasoning_options": [],
2312
3749
  "tool_call": true,
3750
+ "structured_output": true,
2313
3751
  "temperature": true,
2314
- "knowledge": "2025-03-31",
2315
- "release_date": "2025-08-05",
2316
- "last_updated": "2025-08-05",
3752
+ "release_date": "2026-03-11",
3753
+ "last_updated": "2026-03-11",
2317
3754
  "modalities": {
2318
3755
  "input": [
2319
- "text",
2320
- "image",
2321
- "pdf"
3756
+ "text"
2322
3757
  ],
2323
3758
  "output": [
2324
3759
  "text"
2325
3760
  ]
2326
3761
  },
2327
- "open_weights": false,
3762
+ "open_weights": true,
2328
3763
  "limit": {
2329
- "context": 200000,
2330
- "output": 32000
3764
+ "context": 262144,
3765
+ "output": 131072
2331
3766
  },
2332
3767
  "cost": {
2333
- "input": 15,
2334
- "output": 75,
2335
- "cache_read": 1.5,
2336
- "cache_write": 18.75
3768
+ "input": 0.15,
3769
+ "output": 0.65
2337
3770
  }
2338
3771
  },
2339
- "us.anthropic.claude-sonnet-4-6": {
2340
- "id": "us.anthropic.claude-sonnet-4-6",
2341
- "name": "Claude Sonnet 4.6 (US)",
3772
+ "eu.anthropic.claude-sonnet-5": {
3773
+ "id": "eu.anthropic.claude-sonnet-5",
3774
+ "name": "Claude Sonnet 5 (EU)",
3775
+ "description": "Everyday Claude agent model for coding, planning, browsing, and general work",
2342
3776
  "family": "claude-sonnet",
2343
3777
  "attachment": true,
2344
3778
  "reasoning": true,
3779
+ "reasoning_options": [
3780
+ {
3781
+ "type": "toggle"
3782
+ },
3783
+ {
3784
+ "type": "effort",
3785
+ "values": [
3786
+ "low",
3787
+ "medium",
3788
+ "high",
3789
+ "xhigh",
3790
+ "max"
3791
+ ]
3792
+ }
3793
+ ],
2345
3794
  "tool_call": true,
2346
3795
  "structured_output": true,
2347
- "temperature": true,
2348
- "knowledge": "2025-08-31",
2349
- "release_date": "2026-02-17",
2350
- "last_updated": "2026-03-13",
3796
+ "temperature": false,
3797
+ "knowledge": "2026-01-31",
3798
+ "release_date": "2026-06-30",
3799
+ "last_updated": "2026-06-30",
2351
3800
  "modalities": {
2352
3801
  "input": [
2353
3802
  "text",
@@ -2361,61 +3810,73 @@
2361
3810
  "open_weights": false,
2362
3811
  "limit": {
2363
3812
  "context": 1000000,
2364
- "output": 64000
3813
+ "output": 128000
2365
3814
  },
2366
3815
  "cost": {
2367
- "input": 3,
2368
- "output": 15,
2369
- "cache_read": 0.3,
2370
- "cache_write": 3.75
3816
+ "input": 2.2,
3817
+ "output": 11,
3818
+ "cache_read": 0.22,
3819
+ "cache_write": 2.75
2371
3820
  }
2372
3821
  },
2373
- "anthropic.claude-opus-4-6-v1": {
2374
- "id": "anthropic.claude-opus-4-6-v1",
2375
- "name": "Claude Opus 4.6",
2376
- "family": "claude-opus",
2377
- "attachment": true,
3822
+ "nvidia.nemotron-nano-3-30b": {
3823
+ "id": "nvidia.nemotron-nano-3-30b",
3824
+ "name": "NVIDIA Nemotron Nano 3 30B",
3825
+ "description": "Small Nemotron 3 MoE for efficient coding, math, and long-context agents",
3826
+ "family": "nemotron",
3827
+ "attachment": false,
2378
3828
  "reasoning": true,
3829
+ "reasoning_options": [],
2379
3830
  "tool_call": true,
2380
3831
  "structured_output": true,
2381
3832
  "temperature": true,
2382
- "knowledge": "2025-05-31",
2383
- "release_date": "2026-02-05",
2384
- "last_updated": "2026-03-13",
3833
+ "release_date": "2025-12-23",
3834
+ "last_updated": "2025-12-23",
2385
3835
  "modalities": {
2386
3836
  "input": [
2387
- "text",
2388
- "image",
2389
- "pdf"
3837
+ "text"
2390
3838
  ],
2391
3839
  "output": [
2392
3840
  "text"
2393
3841
  ]
2394
3842
  },
2395
- "open_weights": false,
3843
+ "open_weights": true,
2396
3844
  "limit": {
2397
- "context": 1000000,
2398
- "output": 128000
3845
+ "context": 128000,
3846
+ "output": 4096
2399
3847
  },
2400
3848
  "cost": {
2401
- "input": 5,
2402
- "output": 25,
2403
- "cache_read": 0.5,
2404
- "cache_write": 6.25
3849
+ "input": 0.06,
3850
+ "output": 0.24
2405
3851
  }
2406
3852
  },
2407
- "eu.anthropic.claude-sonnet-4-6": {
2408
- "id": "eu.anthropic.claude-sonnet-4-6",
2409
- "name": "Claude Sonnet 4.6 (EU)",
2410
- "family": "claude-sonnet",
3853
+ "eu.anthropic.claude-opus-4-5-20251101-v1:0": {
3854
+ "id": "eu.anthropic.claude-opus-4-5-20251101-v1:0",
3855
+ "name": "Claude Opus 4.5 (EU)",
3856
+ "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
3857
+ "family": "claude-opus",
2411
3858
  "attachment": true,
2412
3859
  "reasoning": true,
3860
+ "reasoning_options": [
3861
+ {
3862
+ "type": "effort",
3863
+ "values": [
3864
+ "low",
3865
+ "medium",
3866
+ "high"
3867
+ ]
3868
+ },
3869
+ {
3870
+ "type": "budget_tokens",
3871
+ "min": 1024
3872
+ }
3873
+ ],
2413
3874
  "tool_call": true,
2414
3875
  "structured_output": true,
2415
3876
  "temperature": true,
2416
- "knowledge": "2025-08-31",
2417
- "release_date": "2026-02-17",
2418
- "last_updated": "2026-03-13",
3877
+ "knowledge": "2025-03-31",
3878
+ "release_date": "2025-11-24",
3879
+ "last_updated": "2025-08-01",
2419
3880
  "modalities": {
2420
3881
  "input": [
2421
3882
  "text",
@@ -2428,22 +3889,38 @@
2428
3889
  },
2429
3890
  "open_weights": false,
2430
3891
  "limit": {
2431
- "context": 1000000,
3892
+ "context": 200000,
2432
3893
  "output": 64000
2433
3894
  },
2434
3895
  "cost": {
2435
- "input": 3,
2436
- "output": 15,
2437
- "cache_read": 0.3,
2438
- "cache_write": 3.75
3896
+ "input": 5.5,
3897
+ "output": 27.5,
3898
+ "cache_read": 0.55,
3899
+ "cache_write": 6.875
2439
3900
  }
2440
3901
  },
2441
- "eu.anthropic.claude-opus-4-6-v1": {
2442
- "id": "eu.anthropic.claude-opus-4-6-v1",
2443
- "name": "Claude Opus 4.6 (EU)",
3902
+ "global.anthropic.claude-opus-4-6-v1": {
3903
+ "id": "global.anthropic.claude-opus-4-6-v1",
3904
+ "name": "Claude Opus 4.6 (Global)",
3905
+ "description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
2444
3906
  "family": "claude-opus",
2445
3907
  "attachment": true,
2446
3908
  "reasoning": true,
3909
+ "reasoning_options": [
3910
+ {
3911
+ "type": "effort",
3912
+ "values": [
3913
+ "low",
3914
+ "medium",
3915
+ "high",
3916
+ "max"
3917
+ ]
3918
+ },
3919
+ {
3920
+ "type": "budget_tokens",
3921
+ "min": 1024
3922
+ }
3923
+ ],
2447
3924
  "tool_call": true,
2448
3925
  "structured_output": true,
2449
3926
  "temperature": true,
@@ -2472,18 +3949,25 @@
2472
3949
  "cache_write": 6.25
2473
3950
  }
2474
3951
  },
2475
- "au.anthropic.claude-haiku-4-5-20251001-v1:0": {
2476
- "id": "au.anthropic.claude-haiku-4-5-20251001-v1:0",
2477
- "name": "Claude Haiku 4.5 (AU)",
2478
- "family": "claude-haiku",
3952
+ "us.anthropic.claude-sonnet-4-5-20250929-v1:0": {
3953
+ "id": "us.anthropic.claude-sonnet-4-5-20250929-v1:0",
3954
+ "name": "Claude Sonnet 4.5 (US)",
3955
+ "description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
3956
+ "family": "claude-sonnet",
2479
3957
  "attachment": true,
2480
3958
  "reasoning": true,
3959
+ "reasoning_options": [
3960
+ {
3961
+ "type": "budget_tokens",
3962
+ "min": 1024
3963
+ }
3964
+ ],
2481
3965
  "tool_call": true,
2482
3966
  "structured_output": true,
2483
3967
  "temperature": true,
2484
- "knowledge": "2025-02-28",
2485
- "release_date": "2025-10-15",
2486
- "last_updated": "2025-10-15",
3968
+ "knowledge": "2025-07-31",
3969
+ "release_date": "2025-09-29",
3970
+ "last_updated": "2025-09-29",
2487
3971
  "modalities": {
2488
3972
  "input": [
2489
3973
  "text",
@@ -2500,29 +3984,29 @@
2500
3984
  "output": 64000
2501
3985
  },
2502
3986
  "cost": {
2503
- "input": 1,
2504
- "output": 5,
2505
- "cache_read": 0.1,
2506
- "cache_write": 1.25
3987
+ "input": 3,
3988
+ "output": 15,
3989
+ "cache_read": 0.3,
3990
+ "cache_write": 3.75
2507
3991
  }
2508
3992
  },
2509
- "jp.anthropic.claude-sonnet-4-5-20250929-v1:0": {
2510
- "id": "jp.anthropic.claude-sonnet-4-5-20250929-v1:0",
2511
- "name": "Claude Sonnet 4.5 (JP)",
2512
- "family": "claude-sonnet",
2513
- "attachment": true,
2514
- "reasoning": true,
2515
- "tool_call": true,
3993
+ "google.gemma-3-12b-it": {
3994
+ "id": "google.gemma-3-12b-it",
3995
+ "name": "Google Gemma 3 12B",
3996
+ "description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
3997
+ "family": "gemma",
3998
+ "attachment": false,
3999
+ "reasoning": false,
4000
+ "tool_call": false,
2516
4001
  "structured_output": true,
2517
4002
  "temperature": true,
2518
- "knowledge": "2025-07-31",
2519
- "release_date": "2025-09-29",
2520
- "last_updated": "2025-09-29",
4003
+ "knowledge": "2024-12",
4004
+ "release_date": "2024-12-01",
4005
+ "last_updated": "2024-12-01",
2521
4006
  "modalities": {
2522
4007
  "input": [
2523
4008
  "text",
2524
- "image",
2525
- "pdf"
4009
+ "image"
2526
4010
  ],
2527
4011
  "output": [
2528
4012
  "text"
@@ -2530,62 +4014,69 @@
2530
4014
  },
2531
4015
  "open_weights": false,
2532
4016
  "limit": {
2533
- "context": 200000,
2534
- "output": 64000
4017
+ "context": 131072,
4018
+ "output": 8192
2535
4019
  },
2536
4020
  "cost": {
2537
- "input": 3,
2538
- "output": 15,
2539
- "cache_read": 0.3,
2540
- "cache_write": 3.75
4021
+ "input": 0.049999999999999996,
4022
+ "output": 0.09999999999999999
2541
4023
  }
2542
4024
  },
2543
- "jp.anthropic.claude-sonnet-4-6": {
2544
- "id": "jp.anthropic.claude-sonnet-4-6",
2545
- "name": "Claude Sonnet 4.6 (JP)",
2546
- "family": "claude-sonnet",
2547
- "attachment": true,
4025
+ "minimax.minimax-m2.1": {
4026
+ "id": "minimax.minimax-m2.1",
4027
+ "name": "MiniMax M2.1",
4028
+ "description": "MiniMax model for chat, coding, office work, and agentic tasks",
4029
+ "family": "minimax",
4030
+ "attachment": false,
2548
4031
  "reasoning": true,
4032
+ "reasoning_options": [],
2549
4033
  "tool_call": true,
2550
- "structured_output": true,
4034
+ "structured_output": false,
2551
4035
  "temperature": true,
2552
- "knowledge": "2025-08-31",
2553
- "release_date": "2026-02-17",
2554
- "last_updated": "2026-03-13",
4036
+ "release_date": "2025-12-23",
4037
+ "last_updated": "2025-12-23",
2555
4038
  "modalities": {
2556
4039
  "input": [
2557
- "text",
2558
- "image",
2559
- "pdf"
4040
+ "text"
2560
4041
  ],
2561
4042
  "output": [
2562
4043
  "text"
2563
4044
  ]
2564
4045
  },
2565
- "open_weights": false,
4046
+ "open_weights": true,
2566
4047
  "limit": {
2567
- "context": 1000000,
2568
- "output": 64000
4048
+ "context": 204800,
4049
+ "output": 131072
2569
4050
  },
2570
4051
  "cost": {
2571
- "input": 3,
2572
- "output": 15,
2573
- "cache_read": 0.3,
2574
- "cache_write": 3.75
4052
+ "input": 0.3,
4053
+ "output": 1.2
2575
4054
  }
2576
4055
  },
2577
- "us.anthropic.claude-opus-4-6-v1": {
2578
- "id": "us.anthropic.claude-opus-4-6-v1",
2579
- "name": "Claude Opus 4.6 (US)",
2580
- "family": "claude-opus",
4056
+ "eu.anthropic.claude-fable-5": {
4057
+ "id": "eu.anthropic.claude-fable-5",
4058
+ "name": "Claude Fable 5 (EU)",
4059
+ "description": "Claude model for creative writing, analysis, and controlled agent workflows",
4060
+ "family": "claude-fable",
2581
4061
  "attachment": true,
2582
4062
  "reasoning": true,
4063
+ "reasoning_options": [
4064
+ {
4065
+ "type": "effort",
4066
+ "values": [
4067
+ "low",
4068
+ "medium",
4069
+ "high",
4070
+ "xhigh",
4071
+ "max"
4072
+ ]
4073
+ }
4074
+ ],
2583
4075
  "tool_call": true,
2584
- "structured_output": true,
2585
- "temperature": true,
2586
- "knowledge": "2025-05-31",
2587
- "release_date": "2026-02-05",
2588
- "last_updated": "2026-03-13",
4076
+ "temperature": false,
4077
+ "knowledge": "2026-01-31",
4078
+ "release_date": "2026-06-09",
4079
+ "last_updated": "2026-06-09",
2589
4080
  "modalities": {
2590
4081
  "input": [
2591
4082
  "text",
@@ -2602,27 +4093,29 @@
2602
4093
  "output": 128000
2603
4094
  },
2604
4095
  "cost": {
2605
- "input": 5,
2606
- "output": 25,
2607
- "cache_read": 0.5,
2608
- "cache_write": 6.25
4096
+ "input": 11,
4097
+ "output": 55,
4098
+ "cache_read": 1.1,
4099
+ "cache_write": 13.75
2609
4100
  }
2610
4101
  },
2611
- "us.meta.llama4-scout-17b-instruct-v1:0": {
2612
- "id": "us.meta.llama4-scout-17b-instruct-v1:0",
2613
- "name": "Llama 4 Scout 17B Instruct (US)",
2614
- "family": "llama",
2615
- "attachment": true,
2616
- "reasoning": false,
4102
+ "deepseek.v3.2": {
4103
+ "id": "deepseek.v3.2",
4104
+ "name": "DeepSeek-V3.2",
4105
+ "description": "DeepSeek chat model for instruction following, coding, and analysis",
4106
+ "family": "deepseek",
4107
+ "attachment": false,
4108
+ "reasoning": true,
4109
+ "reasoning_options": [],
2617
4110
  "tool_call": true,
4111
+ "structured_output": true,
2618
4112
  "temperature": true,
2619
- "knowledge": "2024-08",
2620
- "release_date": "2025-04-05",
2621
- "last_updated": "2025-04-05",
4113
+ "knowledge": "2024-07",
4114
+ "release_date": "2026-02-06",
4115
+ "last_updated": "2026-02-06",
2622
4116
  "modalities": {
2623
4117
  "input": [
2624
- "text",
2625
- "image"
4118
+ "text"
2626
4119
  ],
2627
4120
  "output": [
2628
4121
  "text"
@@ -2630,30 +4123,29 @@
2630
4123
  },
2631
4124
  "open_weights": true,
2632
4125
  "limit": {
2633
- "context": 3500000,
2634
- "output": 16384
4126
+ "context": 163840,
4127
+ "output": 81920
2635
4128
  },
2636
4129
  "cost": {
2637
- "input": 0.17,
2638
- "output": 0.66
4130
+ "input": 0.62,
4131
+ "output": 1.85
2639
4132
  }
2640
4133
  },
2641
- "jp.anthropic.claude-opus-4-7": {
2642
- "id": "jp.anthropic.claude-opus-4-7",
2643
- "name": "Claude Opus 4.7 (JP)",
2644
- "family": "claude-opus",
2645
- "attachment": true,
2646
- "reasoning": true,
4134
+ "mistral.ministral-3-14b-instruct": {
4135
+ "id": "mistral.ministral-3-14b-instruct",
4136
+ "name": "Ministral 14B 3.0",
4137
+ "description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
4138
+ "family": "ministral",
4139
+ "attachment": false,
4140
+ "reasoning": false,
2647
4141
  "tool_call": true,
2648
- "temperature": false,
2649
- "knowledge": "2026-01-31",
2650
- "release_date": "2026-04-16",
2651
- "last_updated": "2026-04-16",
4142
+ "structured_output": true,
4143
+ "temperature": true,
4144
+ "release_date": "2024-12-01",
4145
+ "last_updated": "2024-12-01",
2652
4146
  "modalities": {
2653
4147
  "input": [
2654
- "text",
2655
- "image",
2656
- "pdf"
4148
+ "text"
2657
4149
  ],
2658
4150
  "output": [
2659
4151
  "text"
@@ -2661,14 +4153,12 @@
2661
4153
  },
2662
4154
  "open_weights": false,
2663
4155
  "limit": {
2664
- "context": 1000000,
2665
- "output": 128000
4156
+ "context": 128000,
4157
+ "output": 4096
2666
4158
  },
2667
4159
  "cost": {
2668
- "input": 5,
2669
- "output": 25,
2670
- "cache_read": 0.5,
2671
- "cache_write": 6.25
4160
+ "input": 0.2,
4161
+ "output": 0.2
2672
4162
  }
2673
4163
  }
2674
4164
  }