llm.rb 13.0.0 → 14.0.0

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 (106) hide show
  1. checksums.yaml +4 -4
  2. data/CHANGELOG.md +505 -14
  3. data/README.md +484 -50
  4. data/bin/llm.rb +148 -0
  5. data/data/anthropic.json +206 -263
  6. data/data/bedrock.json +2138 -1860
  7. data/data/deepinfra.json +1003 -624
  8. data/data/deepseek.json +38 -34
  9. data/data/google.json +1079 -371
  10. data/data/mistral.json +448 -368
  11. data/data/moonshot.json +384 -0
  12. data/data/openai.json +974 -1343
  13. data/data/xai.json +154 -126
  14. data/data/zai.json +191 -191
  15. data/lib/llm/agent.rb +123 -20
  16. data/lib/llm/context.rb +71 -88
  17. data/lib/llm/cost.rb +23 -17
  18. data/lib/llm/error.rb +0 -8
  19. data/lib/llm/function/array.rb +3 -3
  20. data/lib/llm/function/async/task.rb +2 -0
  21. data/lib/llm/function/fiber/task.rb +2 -0
  22. data/lib/llm/function/fork/task.rb +2 -0
  23. data/lib/llm/function/ractor/task.rb +2 -0
  24. data/lib/llm/function/sequential/group.rb +4 -1
  25. data/lib/llm/function/sequential/task.rb +1 -1
  26. data/lib/llm/function/task.rb +4 -0
  27. data/lib/llm/function/thread/task.rb +2 -0
  28. data/lib/llm/function.rb +33 -6
  29. data/lib/llm/guard/loop.rb +89 -0
  30. data/lib/llm/guard/null.rb +19 -0
  31. data/lib/llm/guard.rb +61 -0
  32. data/lib/llm/provider.rb +36 -0
  33. data/lib/llm/providers/anthropic/stream_parser.rb +1 -0
  34. data/lib/llm/providers/anthropic.rb +2 -9
  35. data/lib/llm/providers/bedrock/request_adapter.rb +1 -1
  36. data/lib/llm/providers/bedrock/stream_parser.rb +1 -0
  37. data/lib/llm/providers/bedrock.rb +1 -8
  38. data/lib/llm/providers/google/stream_parser.rb +1 -0
  39. data/lib/llm/providers/google.rb +1 -8
  40. data/lib/llm/providers/mistral.rb +1 -1
  41. data/lib/llm/providers/moonshot.rb +76 -0
  42. data/lib/llm/providers/ollama.rb +2 -9
  43. data/lib/llm/providers/openai/responses/stream_parser.rb +1 -0
  44. data/lib/llm/providers/openai/responses.rb +7 -9
  45. data/lib/llm/providers/openai/stream_parser.rb +1 -0
  46. data/lib/llm/providers/openai.rb +4 -11
  47. data/lib/llm/repl/bar.rb +4 -3
  48. data/lib/llm/repl/{transcript.rb → buffer.rb} +69 -29
  49. data/lib/llm/repl/color.rb +78 -0
  50. data/lib/llm/repl/command.rb +12 -5
  51. data/lib/llm/repl/commands/compact.rb +2 -2
  52. data/lib/llm/repl/commands/help.rb +3 -5
  53. data/lib/llm/repl/input/char.rb +46 -0
  54. data/lib/llm/repl/input/row.rb +39 -0
  55. data/lib/llm/repl/input.rb +251 -66
  56. data/lib/llm/repl/markdown/table.rb +11 -3
  57. data/lib/llm/repl/markdown.rb +34 -8
  58. data/lib/llm/repl/node.rb +37 -0
  59. data/lib/llm/repl/status.rb +42 -7
  60. data/lib/llm/repl/stream.rb +18 -6
  61. data/lib/llm/repl/walker.rb +3 -2
  62. data/lib/llm/repl/window.rb +54 -35
  63. data/lib/llm/repl.rb +74 -32
  64. data/lib/llm/skill.rb +20 -4
  65. data/lib/llm/stream.rb +8 -7
  66. data/lib/llm/tool.rb +29 -0
  67. data/lib/llm/tools/{swap_text.rb → edit-file.rb} +3 -3
  68. data/lib/llm/tools/git.rb +3 -0
  69. data/lib/llm/tools/mkdir.rb +3 -0
  70. data/lib/llm/tools/rg.rb +3 -0
  71. data/lib/llm/tools/ruby.rb +46 -0
  72. data/lib/llm/tools/shell.rb +3 -0
  73. data/lib/llm/tracer/pretty_logger.rb +127 -0
  74. data/lib/llm/tracer.rb +1 -0
  75. data/lib/llm/transformer/null.rb +21 -0
  76. data/lib/llm/transformer.rb +55 -0
  77. data/lib/llm/version.rb +1 -1
  78. data/lib/llm.rb +12 -2
  79. data/llm.gemspec +9 -2
  80. data/resources/deepdive/advanced/cancellation.md +74 -0
  81. data/resources/deepdive/advanced/compaction.md +83 -0
  82. data/resources/deepdive/advanced/context.md +267 -0
  83. data/resources/deepdive/advanced/guard.md +371 -0
  84. data/resources/deepdive/advanced/tracer.md +180 -0
  85. data/resources/deepdive/advanced/transformer.md +67 -0
  86. data/resources/deepdive/advanced/transports.md +45 -0
  87. data/resources/deepdive/everything_else/audio.md +122 -0
  88. data/resources/deepdive/everything_else/cost.md +99 -0
  89. data/resources/deepdive/everything_else/images.md +89 -0
  90. data/resources/deepdive/everything_else/object.md +108 -0
  91. data/resources/deepdive/everything_else/ocr.md +48 -0
  92. data/resources/deepdive/fundamentals/agents.md +202 -0
  93. data/resources/deepdive/fundamentals/builtin_tools.md +191 -0
  94. data/resources/deepdive/fundamentals/concurrency.md +104 -0
  95. data/resources/deepdive/fundamentals/database.md +449 -0
  96. data/resources/deepdive/fundamentals/embeddings.md +157 -0
  97. data/resources/deepdive/fundamentals/repl.md +87 -0
  98. data/resources/deepdive/fundamentals/schema.md +61 -0
  99. data/resources/deepdive/fundamentals/skills.md +106 -0
  100. data/resources/deepdive/fundamentals/stream.md +110 -0
  101. data/resources/deepdive/fundamentals/tools.md +265 -0
  102. data/resources/deepdive/protocols/a2a.md +106 -0
  103. data/resources/deepdive/protocols/mcp.md +111 -0
  104. data/resources/deepdive.md +58 -1792
  105. metadata +51 -7
  106. data/lib/llm/loop_guard.rb +0 -107
data/data/openai.json CHANGED
@@ -7,29 +7,61 @@
7
7
  "name": "OpenAI",
8
8
  "doc": "https://platform.openai.com/docs/models",
9
9
  "models": {
10
- "gpt-5-codex": {
11
- "id": "gpt-5-codex",
12
- "name": "GPT-5-Codex",
13
- "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
14
- "family": "gpt-codex",
15
- "attachment": false,
10
+ "gpt-image-2": {
11
+ "id": "gpt-image-2",
12
+ "name": "gpt-image-2",
13
+ "description": "Image model for prompt-driven generation, editing, and visual design workflows",
14
+ "family": "gpt-image",
15
+ "attachment": true,
16
+ "reasoning": false,
17
+ "tool_call": false,
18
+ "temperature": false,
19
+ "release_date": "2026-04-21",
20
+ "last_updated": "2026-04-21",
21
+ "modalities": {
22
+ "input": [
23
+ "text",
24
+ "image"
25
+ ],
26
+ "output": [
27
+ "image"
28
+ ]
29
+ },
30
+ "open_weights": false,
31
+ "limit": {
32
+ "context": 0,
33
+ "input": 0,
34
+ "output": 0
35
+ },
36
+ "cost": {
37
+ "input": 5,
38
+ "output": 30,
39
+ "cache_read": 1.25
40
+ }
41
+ },
42
+ "gpt-5.2-pro": {
43
+ "id": "gpt-5.2-pro",
44
+ "name": "GPT-5.2 Pro",
45
+ "description": "Higher-accuracy GPT-5.2 variant for tougher reasoning and review workflows",
46
+ "family": "gpt-pro",
47
+ "attachment": true,
16
48
  "reasoning": true,
17
49
  "reasoning_options": [
18
50
  {
19
51
  "type": "effort",
20
52
  "values": [
21
- "low",
22
53
  "medium",
23
- "high"
54
+ "high",
55
+ "xhigh"
24
56
  ]
25
57
  }
26
58
  ],
27
59
  "tool_call": true,
28
- "structured_output": true,
60
+ "structured_output": false,
29
61
  "temperature": false,
30
- "knowledge": "2024-09-30",
31
- "release_date": "2025-09-15",
32
- "last_updated": "2025-09-15",
62
+ "knowledge": "2025-08-31",
63
+ "release_date": "2025-12-11",
64
+ "last_updated": "2025-12-11",
33
65
  "modalities": {
34
66
  "input": [
35
67
  "text",
@@ -46,9 +78,8 @@
46
78
  "output": 128000
47
79
  },
48
80
  "cost": {
49
- "input": 1.25,
50
- "output": 10,
51
- "cache_read": 0.125
81
+ "input": 21,
82
+ "output": 168
52
83
  }
53
84
  },
54
85
  "gpt-5.5-pro": {
@@ -109,29 +140,19 @@
109
140
  }
110
141
  }
111
142
  },
112
- "o3": {
113
- "id": "o3",
114
- "name": "o3",
115
- "description": "Deliberate o-series reasoner for hard math, coding, and multi-step analysis",
116
- "family": "o",
143
+ "gpt-4.1-mini": {
144
+ "id": "gpt-4.1-mini",
145
+ "name": "GPT-4.1 mini",
146
+ "description": "Affordable GPT-4.1 lane for fast coding help and structured extraction",
147
+ "family": "gpt-mini",
117
148
  "attachment": true,
118
- "reasoning": true,
119
- "reasoning_options": [
120
- {
121
- "type": "effort",
122
- "values": [
123
- "low",
124
- "medium",
125
- "high"
126
- ]
127
- }
128
- ],
149
+ "reasoning": false,
129
150
  "tool_call": true,
130
151
  "structured_output": true,
131
- "temperature": false,
132
- "knowledge": "2024-05",
133
- "release_date": "2025-04-16",
134
- "last_updated": "2025-04-16",
152
+ "temperature": true,
153
+ "knowledge": "2024-04",
154
+ "release_date": "2025-04-14",
155
+ "last_updated": "2025-04-14",
135
156
  "modalities": {
136
157
  "input": [
137
158
  "text",
@@ -144,40 +165,28 @@
144
165
  },
145
166
  "open_weights": false,
146
167
  "limit": {
147
- "context": 200000,
148
- "output": 100000
168
+ "context": 1047576,
169
+ "output": 32768
149
170
  },
150
171
  "cost": {
151
- "input": 2,
152
- "output": 8,
153
- "cache_read": 0.5
172
+ "input": 0.4,
173
+ "output": 1.6,
174
+ "cache_read": 0.1
154
175
  }
155
176
  },
156
- "gpt-5.4": {
157
- "id": "gpt-5.4",
158
- "name": "GPT-5.4",
159
- "description": "Agent-ready GPT for coding and computer-use workflows at a lower cost",
177
+ "gpt-4o": {
178
+ "id": "gpt-4o",
179
+ "name": "GPT-4o",
180
+ "description": "Omni-era GPT for multimodal chat, practical coding, and general assistants",
160
181
  "family": "gpt",
161
182
  "attachment": true,
162
- "reasoning": true,
163
- "reasoning_options": [
164
- {
165
- "type": "effort",
166
- "values": [
167
- "none",
168
- "low",
169
- "medium",
170
- "high",
171
- "xhigh"
172
- ]
173
- }
174
- ],
183
+ "reasoning": false,
175
184
  "tool_call": true,
176
185
  "structured_output": true,
177
- "temperature": false,
178
- "knowledge": "2025-08-31",
179
- "release_date": "2026-03-05",
180
- "last_updated": "2026-03-05",
186
+ "temperature": true,
187
+ "knowledge": "2023-09",
188
+ "release_date": "2024-05-13",
189
+ "last_updated": "2024-08-06",
181
190
  "modalities": {
182
191
  "input": [
183
192
  "text",
@@ -190,72 +199,45 @@
190
199
  },
191
200
  "open_weights": false,
192
201
  "limit": {
193
- "context": 1050000,
194
- "input": 922000,
195
- "output": 128000
196
- },
197
- "experimental": {
198
- "modes": {
199
- "fast": {
200
- "cost": {
201
- "input": 5,
202
- "output": 30,
203
- "cache_read": 0.5
204
- },
205
- "provider": {
206
- "body": {
207
- "service_tier": "priority"
208
- }
209
- }
210
- }
211
- }
202
+ "context": 128000,
203
+ "output": 16384
212
204
  },
213
205
  "cost": {
214
206
  "input": 2.5,
215
- "output": 15,
216
- "cache_read": 0.25,
217
- "tiers": [
218
- {
219
- "input": 5,
220
- "output": 22.5,
221
- "cache_read": 0.5,
222
- "tier": {
223
- "type": "context",
224
- "size": 272000
225
- }
226
- }
227
- ],
228
- "context_over_200k": {
229
- "input": 5,
230
- "output": 22.5,
231
- "cache_read": 0.5
232
- }
207
+ "output": 10,
208
+ "cache_read": 1.25
233
209
  }
234
210
  },
235
- "o4-mini-deep-research": {
236
- "id": "o4-mini-deep-research",
237
- "name": "o4-mini-deep-research",
238
- "description": "Research model for long-horizon investigation, synthesis, and analytical reports",
239
- "family": "o-mini",
211
+ "gpt-5.3-codex-spark": {
212
+ "id": "gpt-5.3-codex-spark",
213
+ "name": "GPT-5.3 Codex Spark",
214
+ "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
215
+ "family": "gpt-codex-spark",
240
216
  "attachment": true,
241
217
  "reasoning": true,
242
218
  "reasoning_options": [
243
219
  {
244
220
  "type": "effort",
245
221
  "values": [
246
- "medium"
222
+ "none",
223
+ "low",
224
+ "medium",
225
+ "high",
226
+ "xhigh"
247
227
  ]
248
228
  }
249
229
  ],
250
230
  "tool_call": true,
231
+ "structured_output": true,
251
232
  "temperature": false,
252
- "knowledge": "2024-05",
253
- "release_date": "2024-06-26",
254
- "last_updated": "2024-06-26",
233
+ "knowledge": "2025-08-31",
234
+ "release_date": "2026-02-05",
235
+ "last_updated": "2026-02-05",
255
236
  "modalities": {
256
237
  "input": [
257
238
  "text",
258
- "image"
239
+ "image",
240
+ "pdf"
259
241
  ],
260
242
  "output": [
261
243
  "text"
@@ -263,13 +245,14 @@
263
245
  },
264
246
  "open_weights": false,
265
247
  "limit": {
266
- "context": 200000,
267
- "output": 100000
248
+ "context": 128000,
249
+ "input": 100000,
250
+ "output": 32000
268
251
  },
269
252
  "cost": {
270
- "input": 2,
271
- "output": 8,
272
- "cache_read": 0.5
253
+ "input": 1.75,
254
+ "output": 14,
255
+ "cache_read": 0.175
273
256
  }
274
257
  },
275
258
  "gpt-5.4-pro": {
@@ -329,103 +312,142 @@
329
312
  }
330
313
  }
331
314
  },
332
- "gpt-4.1-mini": {
333
- "id": "gpt-4.1-mini",
334
- "name": "GPT-4.1 mini",
335
- "description": "Affordable GPT-4.1 lane for fast coding help and structured extraction",
336
- "family": "gpt-mini",
337
- "attachment": true,
338
- "reasoning": false,
339
- "tool_call": true,
340
- "structured_output": true,
341
- "temperature": true,
342
- "knowledge": "2024-04",
343
- "release_date": "2025-04-14",
344
- "last_updated": "2025-04-14",
345
- "modalities": {
346
- "input": [
347
- "text",
348
- "image",
349
- "pdf"
350
- ],
351
- "output": [
352
- "text"
353
- ]
354
- },
355
- "open_weights": false,
356
- "limit": {
357
- "context": 1047576,
358
- "output": 32768
359
- },
360
- "cost": {
361
- "input": 0.4,
362
- "output": 1.6,
363
- "cache_read": 0.1
364
- }
365
- },
366
- "gpt-realtime-2.1": {
367
- "id": "gpt-realtime-2.1",
368
- "name": "GPT-Realtime-2.1",
369
- "description": "Realtime speech-to-speech model with configurable reasoning, tool use, and robust voice-agent behavior",
370
- "family": "gpt",
315
+ "gpt-5.6-sol": {
316
+ "id": "gpt-5.6-sol",
317
+ "name": "GPT-5.6 Sol",
318
+ "description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
319
+ "family": "gpt-sol",
371
320
  "attachment": true,
372
321
  "reasoning": true,
373
322
  "reasoning_options": [
374
323
  {
375
324
  "type": "effort",
376
325
  "values": [
377
- "minimal",
326
+ "none",
378
327
  "low",
379
328
  "medium",
380
329
  "high",
381
- "xhigh"
330
+ "xhigh",
331
+ "max"
382
332
  ]
383
333
  }
384
334
  ],
385
335
  "tool_call": true,
386
- "structured_output": false,
336
+ "structured_output": true,
387
337
  "temperature": false,
388
- "knowledge": "2024-09-30",
389
- "release_date": "2026-07-06",
390
- "last_updated": "2026-07-06",
338
+ "knowledge": "2026-02-16",
339
+ "release_date": "2026-07-09",
340
+ "last_updated": "2026-07-09",
391
341
  "modalities": {
392
342
  "input": [
393
343
  "text",
394
- "audio",
395
- "image"
344
+ "image",
345
+ "pdf"
396
346
  ],
397
347
  "output": [
398
- "text",
399
- "audio"
348
+ "text"
400
349
  ]
401
350
  },
402
351
  "open_weights": false,
403
352
  "limit": {
404
- "context": 128000,
405
- "input": 96000,
406
- "output": 32000
353
+ "context": 1050000,
354
+ "input": 922000,
355
+ "output": 128000
356
+ },
357
+ "experimental": {
358
+ "modes": {
359
+ "fast": {
360
+ "cost": {
361
+ "input": 10,
362
+ "output": 60,
363
+ "cache_read": 1,
364
+ "cache_write": 12.5
365
+ },
366
+ "provider": {
367
+ "body": {
368
+ "service_tier": "priority"
369
+ }
370
+ }
371
+ },
372
+ "pro": {
373
+ "provider": {
374
+ "body": {
375
+ "reasoning": {
376
+ "mode": "pro"
377
+ }
378
+ }
379
+ }
380
+ }
381
+ }
407
382
  },
408
383
  "cost": {
409
- "input": 4,
410
- "output": 24,
411
- "cache_read": 0.4,
412
- "input_audio": 32,
413
- "output_audio": 64
384
+ "input": 5,
385
+ "output": 30,
386
+ "cache_read": 0.5,
387
+ "cache_write": 6.25,
388
+ "tiers": [
389
+ {
390
+ "input": 10,
391
+ "output": 45,
392
+ "cache_read": 1,
393
+ "cache_write": 12.5,
394
+ "tier": {
395
+ "type": "context",
396
+ "size": 272000
397
+ }
398
+ }
399
+ ],
400
+ "context_over_200k": {
401
+ "input": 10,
402
+ "output": 45,
403
+ "cache_read": 1,
404
+ "cache_write": 12.5
405
+ }
414
406
  }
415
407
  },
416
- "gpt-5.3-chat-latest": {
417
- "id": "gpt-5.3-chat-latest",
418
- "name": "GPT-5.3 Chat (latest)",
419
- "description": "Chat-tuned GPT model for conversational assistance, writing, and tool workflows",
420
- "family": "gpt",
408
+ "text-embedding-ada-002": {
409
+ "id": "text-embedding-ada-002",
410
+ "name": "text-embedding-ada-002",
411
+ "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
412
+ "family": "text-embedding",
413
+ "attachment": false,
414
+ "reasoning": false,
415
+ "tool_call": false,
416
+ "temperature": false,
417
+ "knowledge": "2022-12",
418
+ "release_date": "2022-12-15",
419
+ "last_updated": "2022-12-15",
420
+ "modalities": {
421
+ "input": [
422
+ "text"
423
+ ],
424
+ "output": [
425
+ "text"
426
+ ]
427
+ },
428
+ "open_weights": false,
429
+ "limit": {
430
+ "context": 8192,
431
+ "output": 1536
432
+ },
433
+ "cost": {
434
+ "input": 0.1,
435
+ "output": 0
436
+ }
437
+ },
438
+ "gpt-4.1-nano": {
439
+ "id": "gpt-4.1-nano",
440
+ "name": "GPT-4.1 nano",
441
+ "description": "Tiny GPT-4.1 option for classification, routing, and very high-volume tasks",
442
+ "family": "gpt-nano",
421
443
  "attachment": true,
422
444
  "reasoning": false,
423
445
  "tool_call": true,
424
446
  "structured_output": true,
425
447
  "temperature": true,
426
- "knowledge": "2025-08-31",
427
- "release_date": "2026-03-03",
428
- "last_updated": "2026-03-03",
448
+ "knowledge": "2024-04",
449
+ "release_date": "2025-04-14",
450
+ "last_updated": "2025-04-14",
429
451
  "modalities": {
430
452
  "input": [
431
453
  "text",
@@ -437,43 +459,41 @@
437
459
  },
438
460
  "open_weights": false,
439
461
  "limit": {
440
- "context": 128000,
441
- "output": 16384
462
+ "context": 1047576,
463
+ "output": 32768
442
464
  },
465
+ "status": "deprecated",
443
466
  "cost": {
444
- "input": 1.75,
445
- "output": 14,
446
- "cache_read": 0.175
467
+ "input": 0.1,
468
+ "output": 0.4,
469
+ "cache_read": 0.025
447
470
  }
448
471
  },
449
- "o1": {
450
- "id": "o1",
451
- "name": "o1",
452
- "description": "O-series reasoning model for hard analysis, math, coding, and planning",
453
- "family": "o",
472
+ "gpt-5.2-chat-latest": {
473
+ "id": "gpt-5.2-chat-latest",
474
+ "name": "GPT-5.2 Chat",
475
+ "description": "Chat-tuned GPT model for conversational assistance, writing, and tool workflows",
476
+ "family": "gpt-codex",
454
477
  "attachment": true,
455
478
  "reasoning": true,
456
479
  "reasoning_options": [
457
480
  {
458
481
  "type": "effort",
459
482
  "values": [
460
- "low",
461
- "medium",
462
- "high"
483
+ "medium"
463
484
  ]
464
485
  }
465
486
  ],
466
487
  "tool_call": true,
467
488
  "structured_output": true,
468
489
  "temperature": false,
469
- "knowledge": "2023-09",
470
- "release_date": "2024-12-05",
471
- "last_updated": "2024-12-05",
490
+ "knowledge": "2025-08-31",
491
+ "release_date": "2025-12-11",
492
+ "last_updated": "2025-12-11",
472
493
  "modalities": {
473
494
  "input": [
474
495
  "text",
475
- "image",
476
- "pdf"
496
+ "image"
477
497
  ],
478
498
  "output": [
479
499
  "text"
@@ -481,27 +501,26 @@
481
501
  },
482
502
  "open_weights": false,
483
503
  "limit": {
484
- "context": 200000,
485
- "output": 100000
504
+ "context": 128000,
505
+ "output": 16384
486
506
  },
487
507
  "cost": {
488
- "input": 15,
489
- "output": 60,
490
- "cache_read": 7.5
508
+ "input": 1.75,
509
+ "output": 14,
510
+ "cache_read": 0.175
491
511
  }
492
512
  },
493
- "gpt-5-mini": {
494
- "id": "gpt-5-mini",
495
- "name": "GPT-5 Mini",
496
- "description": "Small GPT-5 for responsive agents, coding help, and everyday automation",
497
- "family": "gpt-mini",
498
- "attachment": true,
513
+ "o3-mini": {
514
+ "id": "o3-mini",
515
+ "name": "o3-mini",
516
+ "description": "Smaller o-series reasoner for economical coding, math, and planning tasks",
517
+ "family": "o-mini",
518
+ "attachment": false,
499
519
  "reasoning": true,
500
520
  "reasoning_options": [
501
521
  {
502
522
  "type": "effort",
503
523
  "values": [
504
- "minimal",
505
524
  "low",
506
525
  "medium",
507
526
  "high"
@@ -511,13 +530,12 @@
511
530
  "tool_call": true,
512
531
  "structured_output": true,
513
532
  "temperature": false,
514
- "knowledge": "2024-05-30",
515
- "release_date": "2025-08-07",
516
- "last_updated": "2025-08-07",
533
+ "knowledge": "2024-05",
534
+ "release_date": "2024-12-20",
535
+ "last_updated": "2025-01-29",
517
536
  "modalities": {
518
537
  "input": [
519
- "text",
520
- "image"
538
+ "text"
521
539
  ],
522
540
  "output": [
523
541
  "text"
@@ -525,68 +543,49 @@
525
543
  },
526
544
  "open_weights": false,
527
545
  "limit": {
528
- "context": 400000,
529
- "input": 272000,
530
- "output": 128000
546
+ "context": 200000,
547
+ "output": 100000
531
548
  },
549
+ "status": "deprecated",
532
550
  "cost": {
533
- "input": 0.25,
534
- "output": 2,
535
- "cache_read": 0.025
551
+ "input": 1.1,
552
+ "output": 4.4,
553
+ "cache_read": 0.55
536
554
  }
537
555
  },
538
- "gpt-5.3-codex-spark": {
539
- "id": "gpt-5.3-codex-spark",
540
- "name": "GPT-5.3 Codex Spark",
541
- "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
542
- "family": "gpt-codex-spark",
556
+ "gpt-image-1-mini": {
557
+ "id": "gpt-image-1-mini",
558
+ "name": "gpt-image-1-mini",
559
+ "description": "Image model for prompt-driven generation, editing, and visual design workflows",
560
+ "family": "gpt-image",
543
561
  "attachment": true,
544
- "reasoning": true,
545
- "reasoning_options": [
546
- {
547
- "type": "effort",
548
- "values": [
549
- "none",
550
- "low",
551
- "medium",
552
- "high",
553
- "xhigh"
554
- ]
555
- }
556
- ],
557
- "tool_call": true,
558
- "structured_output": true,
562
+ "reasoning": false,
563
+ "tool_call": false,
559
564
  "temperature": false,
560
- "knowledge": "2025-08-31",
561
- "release_date": "2026-02-05",
562
- "last_updated": "2026-02-05",
565
+ "release_date": "2025-09-26",
566
+ "last_updated": "2025-09-26",
563
567
  "modalities": {
564
568
  "input": [
565
569
  "text",
566
- "image",
567
- "pdf"
570
+ "image"
568
571
  ],
569
572
  "output": [
570
- "text"
573
+ "text",
574
+ "image"
571
575
  ]
572
576
  },
573
577
  "open_weights": false,
574
578
  "limit": {
575
- "context": 128000,
576
- "input": 100000,
577
- "output": 32000
578
- },
579
- "cost": {
580
- "input": 1.75,
581
- "output": 14,
582
- "cache_read": 0.175
579
+ "context": 0,
580
+ "input": 0,
581
+ "output": 0
583
582
  }
584
583
  },
585
- "gpt-5.4-mini": {
586
- "id": "gpt-5.4-mini",
587
- "name": "GPT-5.4 mini",
588
- "description": "Strong small GPT for coding subagents, quick tool use, and high-volume work",
589
- "family": "gpt-mini",
584
+ "gpt-5.5": {
585
+ "id": "gpt-5.5",
586
+ "name": "GPT-5.5",
587
+ "description": "Default frontier GPT for coding, computer use, research, and knowledge work",
588
+ "family": "gpt",
590
589
  "attachment": true,
591
590
  "reasoning": true,
592
591
  "reasoning_options": [
@@ -604,13 +603,14 @@
604
603
  "tool_call": true,
605
604
  "structured_output": true,
606
605
  "temperature": false,
607
- "knowledge": "2025-08-31",
608
- "release_date": "2026-03-17",
609
- "last_updated": "2026-03-17",
606
+ "knowledge": "2025-12-01",
607
+ "release_date": "2026-04-23",
608
+ "last_updated": "2026-04-23",
610
609
  "modalities": {
611
610
  "input": [
612
611
  "text",
613
- "image"
612
+ "image",
613
+ "pdf"
614
614
  ],
615
615
  "output": [
616
616
  "text"
@@ -618,17 +618,17 @@
618
618
  },
619
619
  "open_weights": false,
620
620
  "limit": {
621
- "context": 400000,
622
- "input": 272000,
621
+ "context": 1050000,
622
+ "input": 922000,
623
623
  "output": 128000
624
624
  },
625
625
  "experimental": {
626
626
  "modes": {
627
627
  "fast": {
628
628
  "cost": {
629
- "input": 1.5,
630
- "output": 9,
631
- "cache_read": 0.15
629
+ "input": 12.5,
630
+ "output": 75,
631
+ "cache_read": 1.25
632
632
  },
633
633
  "provider": {
634
634
  "body": {
@@ -639,15 +639,31 @@
639
639
  }
640
640
  },
641
641
  "cost": {
642
- "input": 0.75,
643
- "output": 4.5,
644
- "cache_read": 0.075
642
+ "input": 5,
643
+ "output": 30,
644
+ "cache_read": 0.5,
645
+ "tiers": [
646
+ {
647
+ "input": 10,
648
+ "output": 45,
649
+ "cache_read": 1,
650
+ "tier": {
651
+ "type": "context",
652
+ "size": 272000
653
+ }
654
+ }
655
+ ],
656
+ "context_over_200k": {
657
+ "input": 10,
658
+ "output": 45,
659
+ "cache_read": 1
660
+ }
645
661
  }
646
662
  },
647
- "o1-pro": {
648
- "id": "o1-pro",
649
- "name": "o1-pro",
650
- "description": "O-series reasoning model for hard analysis, math, coding, and planning",
663
+ "o3-pro": {
664
+ "id": "o3-pro",
665
+ "name": "o3-pro",
666
+ "description": "High-effort o3 tier for difficult technical reasoning and careful answers",
651
667
  "family": "o-pro",
652
668
  "attachment": true,
653
669
  "reasoning": true,
@@ -664,9 +680,9 @@
664
680
  "tool_call": true,
665
681
  "structured_output": true,
666
682
  "temperature": false,
667
- "knowledge": "2023-09",
668
- "release_date": "2025-03-19",
669
- "last_updated": "2025-03-19",
683
+ "knowledge": "2024-05",
684
+ "release_date": "2025-06-10",
685
+ "last_updated": "2025-06-10",
670
686
  "modalities": {
671
687
  "input": [
672
688
  "text",
@@ -682,36 +698,23 @@
682
698
  "output": 100000
683
699
  },
684
700
  "cost": {
685
- "input": 150,
686
- "output": 600
701
+ "input": 20,
702
+ "output": 80
687
703
  }
688
704
  },
689
- "gpt-5.6-terra": {
690
- "id": "gpt-5.6-terra",
691
- "name": "GPT-5.6 Terra",
692
- "description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
693
- "family": "gpt-terra",
705
+ "gpt-4o-mini": {
706
+ "id": "gpt-4o-mini",
707
+ "name": "GPT-4o mini",
708
+ "description": "Small omni GPT for cheap multimodal assistance and production-scale traffic",
709
+ "family": "gpt-mini",
694
710
  "attachment": true,
695
- "reasoning": true,
696
- "reasoning_options": [
697
- {
698
- "type": "effort",
699
- "values": [
700
- "none",
701
- "low",
702
- "medium",
703
- "high",
704
- "xhigh",
705
- "max"
706
- ]
707
- }
708
- ],
711
+ "reasoning": false,
709
712
  "tool_call": true,
710
713
  "structured_output": true,
711
- "temperature": false,
712
- "knowledge": "2026-02-16",
713
- "release_date": "2026-07-09",
714
- "last_updated": "2026-07-09",
714
+ "temperature": true,
715
+ "knowledge": "2023-09",
716
+ "release_date": "2024-07-18",
717
+ "last_updated": "2024-07-18",
715
718
  "modalities": {
716
719
  "input": [
717
720
  "text",
@@ -724,72 +727,26 @@
724
727
  },
725
728
  "open_weights": false,
726
729
  "limit": {
727
- "context": 1050000,
728
- "input": 922000,
729
- "output": 128000
730
- },
731
- "experimental": {
732
- "modes": {
733
- "fast": {
734
- "cost": {
735
- "input": 5,
736
- "output": 30,
737
- "cache_read": 0.5,
738
- "cache_write": 6.25
739
- },
740
- "provider": {
741
- "body": {
742
- "service_tier": "priority"
743
- }
744
- }
745
- },
746
- "pro": {
747
- "provider": {
748
- "body": {
749
- "reasoning": {
750
- "mode": "pro"
751
- }
752
- }
753
- }
754
- }
755
- }
730
+ "context": 128000,
731
+ "output": 16384
756
732
  },
757
733
  "cost": {
758
- "input": 2.5,
759
- "output": 15,
760
- "cache_read": 0.25,
761
- "cache_write": 3.125,
762
- "tiers": [
763
- {
764
- "input": 5,
765
- "output": 22.5,
766
- "cache_read": 0.5,
767
- "cache_write": 6.25,
768
- "tier": {
769
- "type": "context",
770
- "size": 272000
771
- }
772
- }
773
- ],
774
- "context_over_200k": {
775
- "input": 5,
776
- "output": 22.5,
777
- "cache_read": 0.5,
778
- "cache_write": 6.25
779
- }
734
+ "input": 0.15,
735
+ "output": 0.6,
736
+ "cache_read": 0.075
780
737
  }
781
738
  },
782
- "gpt-image-1-mini": {
783
- "id": "gpt-image-1-mini",
784
- "name": "gpt-image-1-mini",
739
+ "chatgpt-image-latest": {
740
+ "id": "chatgpt-image-latest",
741
+ "name": "chatgpt-image-latest",
785
742
  "description": "Image model for prompt-driven generation, editing, and visual design workflows",
786
743
  "family": "gpt-image",
787
744
  "attachment": true,
788
745
  "reasoning": false,
789
746
  "tool_call": false,
790
747
  "temperature": false,
791
- "release_date": "2025-09-26",
792
- "last_updated": "2025-09-26",
748
+ "release_date": "2025-12-16",
749
+ "last_updated": "2025-12-16",
793
750
  "modalities": {
794
751
  "input": [
795
752
  "text",
@@ -807,36 +764,34 @@
807
764
  "output": 0
808
765
  }
809
766
  },
810
- "gpt-5.3-codex": {
811
- "id": "gpt-5.3-codex",
812
- "name": "GPT-5.3 Codex",
813
- "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
814
- "family": "gpt-codex",
767
+ "gpt-5": {
768
+ "id": "gpt-5",
769
+ "name": "GPT-5",
770
+ "description": "Original GPT-5 workhorse for reasoning, coding, writing, and tool workflows",
771
+ "family": "gpt",
815
772
  "attachment": true,
816
773
  "reasoning": true,
817
774
  "reasoning_options": [
818
775
  {
819
776
  "type": "effort",
820
777
  "values": [
821
- "none",
778
+ "minimal",
822
779
  "low",
823
780
  "medium",
824
- "high",
825
- "xhigh"
781
+ "high"
826
782
  ]
827
783
  }
828
784
  ],
829
785
  "tool_call": true,
830
786
  "structured_output": true,
831
787
  "temperature": false,
832
- "knowledge": "2025-08-31",
833
- "release_date": "2026-02-05",
834
- "last_updated": "2026-02-05",
788
+ "knowledge": "2024-09-30",
789
+ "release_date": "2025-08-07",
790
+ "last_updated": "2025-08-07",
835
791
  "modalities": {
836
792
  "input": [
837
793
  "text",
838
- "image",
839
- "pdf"
794
+ "image"
840
795
  ],
841
796
  "output": [
842
797
  "text"
@@ -849,22 +804,23 @@
849
804
  "output": 128000
850
805
  },
851
806
  "cost": {
852
- "input": 1.75,
853
- "output": 14,
854
- "cache_read": 0.175
807
+ "input": 1.25,
808
+ "output": 10,
809
+ "cache_read": 0.125
855
810
  }
856
811
  },
857
- "gpt-5.1-codex-max": {
858
- "id": "gpt-5.1-codex-max",
859
- "name": "GPT-5.1 Codex Max",
860
- "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
861
- "family": "gpt-codex",
812
+ "gpt-realtime-2.1": {
813
+ "id": "gpt-realtime-2.1",
814
+ "name": "GPT-Realtime-2.1",
815
+ "description": "Realtime speech-to-speech model with configurable reasoning, tool use, and robust voice-agent behavior",
816
+ "family": "gpt",
862
817
  "attachment": true,
863
818
  "reasoning": true,
864
819
  "reasoning_options": [
865
820
  {
866
821
  "type": "effort",
867
822
  "values": [
823
+ "minimal",
868
824
  "low",
869
825
  "medium",
870
826
  "high",
@@ -873,57 +829,52 @@
873
829
  }
874
830
  ],
875
831
  "tool_call": true,
876
- "structured_output": true,
832
+ "structured_output": false,
877
833
  "temperature": false,
878
834
  "knowledge": "2024-09-30",
879
- "release_date": "2025-11-13",
880
- "last_updated": "2025-11-13",
835
+ "release_date": "2026-07-06",
836
+ "last_updated": "2026-07-06",
881
837
  "modalities": {
882
838
  "input": [
883
839
  "text",
840
+ "audio",
884
841
  "image"
885
842
  ],
886
843
  "output": [
887
- "text"
844
+ "text",
845
+ "audio"
888
846
  ]
889
847
  },
890
848
  "open_weights": false,
891
849
  "limit": {
892
- "context": 400000,
893
- "input": 272000,
894
- "output": 128000
850
+ "context": 128000,
851
+ "input": 96000,
852
+ "output": 32000
895
853
  },
896
854
  "cost": {
897
- "input": 1.25,
898
- "output": 10,
899
- "cache_read": 0.125
855
+ "input": 4,
856
+ "output": 24,
857
+ "cache_read": 0.4,
858
+ "input_audio": 32,
859
+ "output_audio": 64
900
860
  }
901
861
  },
902
- "gpt-5.1-chat-latest": {
903
- "id": "gpt-5.1-chat-latest",
904
- "name": "GPT-5.1 Chat",
905
- "description": "Chat-tuned GPT-5.1 for polished assistants, writing, and product conversations",
906
- "family": "gpt-codex",
907
- "attachment": true,
908
- "reasoning": true,
909
- "reasoning_options": [
910
- {
911
- "type": "effort",
912
- "values": [
913
- "medium"
914
- ]
915
- }
916
- ],
917
- "tool_call": true,
918
- "structured_output": true,
919
- "temperature": false,
920
- "knowledge": "2024-09-30",
921
- "release_date": "2025-11-13",
922
- "last_updated": "2025-11-13",
862
+ "gpt-3.5-turbo": {
863
+ "id": "gpt-3.5-turbo",
864
+ "name": "GPT-3.5-turbo",
865
+ "description": "Compact GPT model for low-latency assistance and high-volume workloads",
866
+ "family": "gpt",
867
+ "attachment": false,
868
+ "reasoning": false,
869
+ "tool_call": false,
870
+ "structured_output": false,
871
+ "temperature": true,
872
+ "knowledge": "2021-09-01",
873
+ "release_date": "2023-03-01",
874
+ "last_updated": "2023-11-06",
923
875
  "modalities": {
924
876
  "input": [
925
- "text",
926
- "image"
877
+ "text"
927
878
  ],
928
879
  "output": [
929
880
  "text"
@@ -931,41 +882,66 @@
931
882
  },
932
883
  "open_weights": false,
933
884
  "limit": {
934
- "context": 128000,
935
- "output": 16384
885
+ "context": 16385,
886
+ "output": 4096
936
887
  },
888
+ "status": "deprecated",
937
889
  "cost": {
938
- "input": 1.25,
939
- "output": 10,
940
- "cache_read": 0.125
890
+ "input": 0.5,
891
+ "output": 1.5,
892
+ "cache_read": 0
941
893
  }
942
894
  },
943
- "o3-mini": {
944
- "id": "o3-mini",
945
- "name": "o3-mini",
946
- "description": "Smaller o-series reasoner for economical coding, math, and planning tasks",
947
- "family": "o-mini",
948
- "attachment": false,
949
- "reasoning": true,
950
- "reasoning_options": [
951
- {
952
- "type": "effort",
953
- "values": [
954
- "low",
955
- "medium",
956
- "high"
957
- ]
958
- }
959
- ],
895
+ "gpt-4o-2024-05-13": {
896
+ "id": "gpt-4o-2024-05-13",
897
+ "name": "GPT-4o (2024-05-13)",
898
+ "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
899
+ "family": "gpt",
900
+ "attachment": true,
901
+ "reasoning": false,
960
902
  "tool_call": true,
961
903
  "structured_output": true,
962
- "temperature": false,
963
- "knowledge": "2024-05",
964
- "release_date": "2024-12-20",
965
- "last_updated": "2025-01-29",
904
+ "temperature": true,
905
+ "knowledge": "2023-09",
906
+ "release_date": "2024-05-13",
907
+ "last_updated": "2024-05-13",
966
908
  "modalities": {
967
909
  "input": [
910
+ "text",
911
+ "image"
912
+ ],
913
+ "output": [
968
914
  "text"
915
+ ]
916
+ },
917
+ "open_weights": false,
918
+ "limit": {
919
+ "context": 128000,
920
+ "output": 4096
921
+ },
922
+ "status": "deprecated",
923
+ "cost": {
924
+ "input": 5,
925
+ "output": 15
926
+ }
927
+ },
928
+ "gpt-4o-2024-11-20": {
929
+ "id": "gpt-4o-2024-11-20",
930
+ "name": "GPT-4o (2024-11-20)",
931
+ "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
932
+ "family": "gpt",
933
+ "attachment": true,
934
+ "reasoning": false,
935
+ "tool_call": true,
936
+ "structured_output": true,
937
+ "temperature": true,
938
+ "knowledge": "2023-09",
939
+ "release_date": "2024-11-20",
940
+ "last_updated": "2024-11-20",
941
+ "modalities": {
942
+ "input": [
943
+ "text",
944
+ "image"
969
945
  ],
970
946
  "output": [
971
947
  "text"
@@ -973,42 +949,45 @@
973
949
  },
974
950
  "open_weights": false,
975
951
  "limit": {
976
- "context": 200000,
977
- "output": 100000
952
+ "context": 128000,
953
+ "output": 16384
978
954
  },
979
955
  "cost": {
980
- "input": 1.1,
981
- "output": 4.4,
982
- "cache_read": 0.55
956
+ "input": 2.5,
957
+ "output": 10,
958
+ "cache_read": 1.25
983
959
  }
984
960
  },
985
- "gpt-5.1-codex": {
986
- "id": "gpt-5.1-codex",
987
- "name": "GPT-5.1 Codex",
988
- "description": "Codex GPT for repository edits, code review, and practical software agents",
989
- "family": "gpt-codex",
961
+ "gpt-5.4": {
962
+ "id": "gpt-5.4",
963
+ "name": "GPT-5.4",
964
+ "description": "Agent-ready GPT for coding and computer-use workflows at a lower cost",
965
+ "family": "gpt",
990
966
  "attachment": true,
991
967
  "reasoning": true,
992
968
  "reasoning_options": [
993
969
  {
994
970
  "type": "effort",
995
971
  "values": [
972
+ "none",
996
973
  "low",
997
974
  "medium",
998
- "high"
975
+ "high",
976
+ "xhigh"
999
977
  ]
1000
978
  }
1001
979
  ],
1002
980
  "tool_call": true,
1003
981
  "structured_output": true,
1004
982
  "temperature": false,
1005
- "knowledge": "2024-09-30",
1006
- "release_date": "2025-11-13",
1007
- "last_updated": "2025-11-13",
983
+ "knowledge": "2025-08-31",
984
+ "release_date": "2026-03-05",
985
+ "last_updated": "2026-03-05",
1008
986
  "modalities": {
1009
987
  "input": [
1010
988
  "text",
1011
- "image"
989
+ "image",
990
+ "pdf"
1012
991
  ],
1013
992
  "output": [
1014
993
  "text"
@@ -1016,32 +995,65 @@
1016
995
  },
1017
996
  "open_weights": false,
1018
997
  "limit": {
1019
- "context": 400000,
1020
- "input": 272000,
998
+ "context": 1050000,
999
+ "input": 922000,
1021
1000
  "output": 128000
1022
1001
  },
1002
+ "experimental": {
1003
+ "modes": {
1004
+ "fast": {
1005
+ "cost": {
1006
+ "input": 5,
1007
+ "output": 30,
1008
+ "cache_read": 0.5
1009
+ },
1010
+ "provider": {
1011
+ "body": {
1012
+ "service_tier": "priority"
1013
+ }
1014
+ }
1015
+ }
1016
+ }
1017
+ },
1023
1018
  "cost": {
1024
- "input": 1.25,
1025
- "output": 10,
1026
- "cache_read": 0.125
1019
+ "input": 2.5,
1020
+ "output": 15,
1021
+ "cache_read": 0.25,
1022
+ "tiers": [
1023
+ {
1024
+ "input": 5,
1025
+ "output": 22.5,
1026
+ "cache_read": 0.5,
1027
+ "tier": {
1028
+ "type": "context",
1029
+ "size": 272000
1030
+ }
1031
+ }
1032
+ ],
1033
+ "context_over_200k": {
1034
+ "input": 5,
1035
+ "output": 22.5,
1036
+ "cache_read": 0.5
1037
+ }
1027
1038
  }
1028
1039
  },
1029
- "gpt-4": {
1030
- "id": "gpt-4",
1031
- "name": "GPT-4",
1032
- "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
1040
+ "gpt-5.3-chat-latest": {
1041
+ "id": "gpt-5.3-chat-latest",
1042
+ "name": "GPT-5.3 Chat (latest)",
1043
+ "description": "Chat-tuned GPT model for conversational assistance, writing, and tool workflows",
1033
1044
  "family": "gpt",
1034
1045
  "attachment": true,
1035
1046
  "reasoning": false,
1036
1047
  "tool_call": true,
1037
- "structured_output": false,
1048
+ "structured_output": true,
1038
1049
  "temperature": true,
1039
- "knowledge": "2023-11",
1040
- "release_date": "2023-11-06",
1041
- "last_updated": "2024-04-09",
1050
+ "knowledge": "2025-08-31",
1051
+ "release_date": "2026-03-03",
1052
+ "last_updated": "2026-03-03",
1042
1053
  "modalities": {
1043
1054
  "input": [
1044
- "text"
1055
+ "text",
1056
+ "image"
1045
1057
  ],
1046
1058
  "output": [
1047
1059
  "text"
@@ -1049,12 +1061,41 @@
1049
1061
  },
1050
1062
  "open_weights": false,
1051
1063
  "limit": {
1052
- "context": 8192,
1053
- "output": 8192
1064
+ "context": 128000,
1065
+ "output": 16384
1054
1066
  },
1055
1067
  "cost": {
1056
- "input": 30,
1057
- "output": 60
1068
+ "input": 1.75,
1069
+ "output": 14,
1070
+ "cache_read": 0.175
1071
+ }
1072
+ },
1073
+ "gpt-image-1.5": {
1074
+ "id": "gpt-image-1.5",
1075
+ "name": "gpt-image-1.5",
1076
+ "description": "Image model for prompt-driven generation, editing, and visual design workflows",
1077
+ "family": "gpt-image",
1078
+ "attachment": true,
1079
+ "reasoning": false,
1080
+ "tool_call": false,
1081
+ "temperature": false,
1082
+ "release_date": "2025-11-25",
1083
+ "last_updated": "2025-11-25",
1084
+ "modalities": {
1085
+ "input": [
1086
+ "text",
1087
+ "image"
1088
+ ],
1089
+ "output": [
1090
+ "text",
1091
+ "image"
1092
+ ]
1093
+ },
1094
+ "open_weights": false,
1095
+ "limit": {
1096
+ "context": 0,
1097
+ "input": 0,
1098
+ "output": 0
1058
1099
  }
1059
1100
  },
1060
1101
  "gpt-5.4-nano": {
@@ -1103,53 +1144,27 @@
1103
1144
  "cache_read": 0.02
1104
1145
  }
1105
1146
  },
1106
- "gpt-4.1": {
1107
- "id": "gpt-4.1",
1108
- "name": "GPT-4.1",
1109
- "description": "Long-lived GPT workhorse for coding, instruction following, and production apps",
1110
- "family": "gpt",
1111
- "attachment": true,
1112
- "reasoning": false,
1113
- "tool_call": true,
1114
- "structured_output": true,
1115
- "temperature": true,
1116
- "knowledge": "2024-04",
1117
- "release_date": "2025-04-14",
1118
- "last_updated": "2025-04-14",
1119
- "modalities": {
1120
- "input": [
1121
- "text",
1122
- "image",
1123
- "pdf"
1124
- ],
1125
- "output": [
1126
- "text"
1127
- ]
1128
- },
1129
- "open_weights": false,
1130
- "limit": {
1131
- "context": 1047576,
1132
- "output": 32768
1133
- },
1134
- "cost": {
1135
- "input": 2,
1136
- "output": 8,
1137
- "cache_read": 0.5
1138
- }
1139
- },
1140
- "gpt-4o-2024-05-13": {
1141
- "id": "gpt-4o-2024-05-13",
1142
- "name": "GPT-4o (2024-05-13)",
1143
- "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
1144
- "family": "gpt",
1147
+ "gpt-5-pro": {
1148
+ "id": "gpt-5-pro",
1149
+ "name": "GPT-5 Pro",
1150
+ "description": "Higher-accuracy GPT-5 tier for tough analysis, coding reviews, and planning",
1151
+ "family": "gpt-pro",
1145
1152
  "attachment": true,
1146
- "reasoning": false,
1153
+ "reasoning": true,
1154
+ "reasoning_options": [
1155
+ {
1156
+ "type": "effort",
1157
+ "values": [
1158
+ "high"
1159
+ ]
1160
+ }
1161
+ ],
1147
1162
  "tool_call": true,
1148
1163
  "structured_output": true,
1149
- "temperature": true,
1150
- "knowledge": "2023-09",
1151
- "release_date": "2024-05-13",
1152
- "last_updated": "2024-05-13",
1164
+ "temperature": false,
1165
+ "knowledge": "2024-09-30",
1166
+ "release_date": "2025-10-06",
1167
+ "last_updated": "2025-10-06",
1153
1168
  "modalities": {
1154
1169
  "input": [
1155
1170
  "text",
@@ -1161,459 +1176,44 @@
1161
1176
  },
1162
1177
  "open_weights": false,
1163
1178
  "limit": {
1164
- "context": 128000,
1165
- "output": 4096
1179
+ "context": 400000,
1180
+ "input": 272000,
1181
+ "output": 272000
1166
1182
  },
1167
1183
  "cost": {
1168
- "input": 5,
1169
- "output": 15
1184
+ "input": 15,
1185
+ "output": 120
1170
1186
  }
1171
1187
  },
1172
- "gpt-4o-2024-11-20": {
1173
- "id": "gpt-4o-2024-11-20",
1174
- "name": "GPT-4o (2024-11-20)",
1175
- "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
1176
- "family": "gpt",
1177
- "attachment": true,
1178
- "reasoning": false,
1179
- "tool_call": true,
1180
- "structured_output": true,
1181
- "temperature": true,
1182
- "knowledge": "2023-09",
1183
- "release_date": "2024-11-20",
1184
- "last_updated": "2024-11-20",
1185
- "modalities": {
1186
- "input": [
1187
- "text",
1188
- "image"
1189
- ],
1190
- "output": [
1191
- "text"
1192
- ]
1193
- },
1194
- "open_weights": false,
1195
- "limit": {
1196
- "context": 128000,
1197
- "output": 16384
1198
- },
1199
- "cost": {
1200
- "input": 2.5,
1201
- "output": 10,
1202
- "cache_read": 1.25
1203
- }
1204
- },
1205
- "gpt-3.5-turbo": {
1206
- "id": "gpt-3.5-turbo",
1207
- "name": "GPT-3.5-turbo",
1208
- "description": "Compact GPT model for low-latency assistance and high-volume workloads",
1209
- "family": "gpt",
1210
- "attachment": false,
1211
- "reasoning": false,
1212
- "tool_call": false,
1213
- "structured_output": false,
1214
- "temperature": true,
1215
- "knowledge": "2021-09-01",
1216
- "release_date": "2023-03-01",
1217
- "last_updated": "2023-11-06",
1218
- "modalities": {
1219
- "input": [
1220
- "text"
1221
- ],
1222
- "output": [
1223
- "text"
1224
- ]
1225
- },
1226
- "open_weights": false,
1227
- "limit": {
1228
- "context": 16385,
1229
- "output": 4096
1230
- },
1231
- "cost": {
1232
- "input": 0.5,
1233
- "output": 1.5,
1234
- "cache_read": 0
1235
- }
1236
- },
1237
- "gpt-4.1-nano": {
1238
- "id": "gpt-4.1-nano",
1239
- "name": "GPT-4.1 nano",
1240
- "description": "Tiny GPT-4.1 option for classification, routing, and very high-volume tasks",
1241
- "family": "gpt-nano",
1242
- "attachment": true,
1243
- "reasoning": false,
1244
- "tool_call": true,
1245
- "structured_output": true,
1246
- "temperature": true,
1247
- "knowledge": "2024-04",
1248
- "release_date": "2025-04-14",
1249
- "last_updated": "2025-04-14",
1250
- "modalities": {
1251
- "input": [
1252
- "text",
1253
- "image"
1254
- ],
1255
- "output": [
1256
- "text"
1257
- ]
1258
- },
1259
- "open_weights": false,
1260
- "limit": {
1261
- "context": 1047576,
1262
- "output": 32768
1263
- },
1264
- "cost": {
1265
- "input": 0.1,
1266
- "output": 0.4,
1267
- "cache_read": 0.025
1268
- }
1269
- },
1270
- "gpt-image-1.5": {
1271
- "id": "gpt-image-1.5",
1272
- "name": "gpt-image-1.5",
1273
- "description": "Image model for prompt-driven generation, editing, and visual design workflows",
1274
- "family": "gpt-image",
1275
- "attachment": true,
1276
- "reasoning": false,
1277
- "tool_call": false,
1278
- "temperature": false,
1279
- "release_date": "2025-11-25",
1280
- "last_updated": "2025-11-25",
1281
- "modalities": {
1282
- "input": [
1283
- "text",
1284
- "image"
1285
- ],
1286
- "output": [
1287
- "text",
1288
- "image"
1289
- ]
1290
- },
1291
- "open_weights": false,
1292
- "limit": {
1293
- "context": 0,
1294
- "input": 0,
1295
- "output": 0
1296
- }
1297
- },
1298
- "gpt-5.2-codex": {
1299
- "id": "gpt-5.2-codex",
1300
- "name": "GPT-5.2 Codex",
1301
- "description": "Code-specialist GPT for repository edits, reviews, and long-running software agents",
1302
- "family": "gpt-codex",
1303
- "attachment": true,
1304
- "reasoning": true,
1305
- "reasoning_options": [
1306
- {
1307
- "type": "effort",
1308
- "values": [
1309
- "low",
1310
- "medium",
1311
- "high",
1312
- "xhigh"
1313
- ]
1314
- }
1315
- ],
1316
- "tool_call": true,
1317
- "structured_output": true,
1318
- "temperature": false,
1319
- "knowledge": "2025-08-31",
1320
- "release_date": "2025-12-11",
1321
- "last_updated": "2025-12-11",
1322
- "modalities": {
1323
- "input": [
1324
- "text",
1325
- "image",
1326
- "pdf"
1327
- ],
1328
- "output": [
1329
- "text"
1330
- ]
1331
- },
1332
- "open_weights": false,
1333
- "limit": {
1334
- "context": 400000,
1335
- "input": 272000,
1336
- "output": 128000
1337
- },
1338
- "cost": {
1339
- "input": 1.75,
1340
- "output": 14,
1341
- "cache_read": 0.175
1342
- }
1343
- },
1344
- "text-embedding-3-large": {
1345
- "id": "text-embedding-3-large",
1346
- "name": "text-embedding-3-large",
1347
- "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
1348
- "family": "text-embedding",
1349
- "attachment": false,
1350
- "reasoning": false,
1351
- "tool_call": false,
1352
- "temperature": false,
1353
- "knowledge": "2024-01",
1354
- "release_date": "2024-01-25",
1355
- "last_updated": "2024-01-25",
1356
- "modalities": {
1357
- "input": [
1358
- "text"
1359
- ],
1360
- "output": [
1361
- "text"
1362
- ]
1363
- },
1364
- "open_weights": false,
1365
- "limit": {
1366
- "context": 8191,
1367
- "output": 3072
1368
- },
1369
- "cost": {
1370
- "input": 0.13,
1371
- "output": 0
1372
- }
1373
- },
1374
- "gpt-5.1-codex-mini": {
1375
- "id": "gpt-5.1-codex-mini",
1376
- "name": "GPT-5.1 Codex mini",
1377
- "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
1378
- "family": "gpt-codex",
1379
- "attachment": true,
1380
- "reasoning": true,
1381
- "reasoning_options": [
1382
- {
1383
- "type": "effort",
1384
- "values": [
1385
- "low",
1386
- "medium",
1387
- "high"
1388
- ]
1389
- }
1390
- ],
1391
- "tool_call": true,
1392
- "structured_output": true,
1393
- "temperature": false,
1394
- "knowledge": "2024-09-30",
1395
- "release_date": "2025-11-13",
1396
- "last_updated": "2025-11-13",
1397
- "modalities": {
1398
- "input": [
1399
- "text",
1400
- "image"
1401
- ],
1402
- "output": [
1403
- "text"
1404
- ]
1405
- },
1406
- "open_weights": false,
1407
- "limit": {
1408
- "context": 400000,
1409
- "input": 272000,
1410
- "output": 128000
1411
- },
1412
- "cost": {
1413
- "input": 0.25,
1414
- "output": 2,
1415
- "cache_read": 0.025
1416
- }
1417
- },
1418
- "gpt-5.2-chat-latest": {
1419
- "id": "gpt-5.2-chat-latest",
1420
- "name": "GPT-5.2 Chat",
1421
- "description": "Chat-tuned GPT model for conversational assistance, writing, and tool workflows",
1422
- "family": "gpt-codex",
1423
- "attachment": true,
1424
- "reasoning": true,
1425
- "reasoning_options": [
1426
- {
1427
- "type": "effort",
1428
- "values": [
1429
- "medium"
1430
- ]
1431
- }
1432
- ],
1433
- "tool_call": true,
1434
- "structured_output": true,
1435
- "temperature": false,
1436
- "knowledge": "2025-08-31",
1437
- "release_date": "2025-12-11",
1438
- "last_updated": "2025-12-11",
1439
- "modalities": {
1440
- "input": [
1441
- "text",
1442
- "image"
1443
- ],
1444
- "output": [
1445
- "text"
1446
- ]
1447
- },
1448
- "open_weights": false,
1449
- "limit": {
1450
- "context": 128000,
1451
- "output": 16384
1452
- },
1453
- "cost": {
1454
- "input": 1.75,
1455
- "output": 14,
1456
- "cache_read": 0.175
1457
- }
1458
- },
1459
- "gpt-5": {
1460
- "id": "gpt-5",
1461
- "name": "GPT-5",
1462
- "description": "Original GPT-5 workhorse for reasoning, coding, writing, and tool workflows",
1463
- "family": "gpt",
1464
- "attachment": true,
1465
- "reasoning": true,
1466
- "reasoning_options": [
1467
- {
1468
- "type": "effort",
1469
- "values": [
1470
- "minimal",
1471
- "low",
1472
- "medium",
1473
- "high"
1474
- ]
1475
- }
1476
- ],
1477
- "tool_call": true,
1478
- "structured_output": true,
1479
- "temperature": false,
1480
- "knowledge": "2024-09-30",
1481
- "release_date": "2025-08-07",
1482
- "last_updated": "2025-08-07",
1483
- "modalities": {
1484
- "input": [
1485
- "text",
1486
- "image"
1487
- ],
1488
- "output": [
1489
- "text"
1490
- ]
1491
- },
1492
- "open_weights": false,
1493
- "limit": {
1494
- "context": 400000,
1495
- "input": 272000,
1496
- "output": 128000
1497
- },
1498
- "cost": {
1499
- "input": 1.25,
1500
- "output": 10,
1501
- "cache_read": 0.125
1502
- }
1503
- },
1504
- "gpt-5-chat-latest": {
1505
- "id": "gpt-5-chat-latest",
1506
- "name": "GPT-5 Chat (latest)",
1507
- "description": "Chat-tuned GPT model for conversational assistance, writing, and tool workflows",
1508
- "family": "gpt-codex",
1509
- "attachment": true,
1510
- "reasoning": true,
1511
- "reasoning_options": [],
1512
- "tool_call": false,
1513
- "structured_output": true,
1514
- "temperature": true,
1515
- "knowledge": "2024-09-30",
1516
- "release_date": "2025-08-07",
1517
- "last_updated": "2025-08-07",
1518
- "modalities": {
1519
- "input": [
1520
- "text",
1521
- "image"
1522
- ],
1523
- "output": [
1524
- "text"
1525
- ]
1526
- },
1527
- "open_weights": false,
1528
- "limit": {
1529
- "context": 400000,
1530
- "input": 272000,
1531
- "output": 128000
1532
- },
1533
- "cost": {
1534
- "input": 1.25,
1535
- "output": 10,
1536
- "cache_read": 0.125
1537
- }
1538
- },
1539
- "text-embedding-ada-002": {
1540
- "id": "text-embedding-ada-002",
1541
- "name": "text-embedding-ada-002",
1542
- "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
1543
- "family": "text-embedding",
1544
- "attachment": false,
1545
- "reasoning": false,
1546
- "tool_call": false,
1547
- "temperature": false,
1548
- "knowledge": "2022-12",
1549
- "release_date": "2022-12-15",
1550
- "last_updated": "2022-12-15",
1551
- "modalities": {
1552
- "input": [
1553
- "text"
1554
- ],
1555
- "output": [
1556
- "text"
1557
- ]
1558
- },
1559
- "open_weights": false,
1560
- "limit": {
1561
- "context": 8192,
1562
- "output": 1536
1563
- },
1564
- "cost": {
1565
- "input": 0.1,
1566
- "output": 0
1567
- }
1568
- },
1569
- "text-embedding-3-small": {
1570
- "id": "text-embedding-3-small",
1571
- "name": "text-embedding-3-small",
1572
- "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
1573
- "family": "text-embedding",
1574
- "attachment": false,
1575
- "reasoning": false,
1576
- "tool_call": false,
1577
- "temperature": false,
1578
- "knowledge": "2024-01",
1579
- "release_date": "2024-01-25",
1580
- "last_updated": "2024-01-25",
1581
- "modalities": {
1582
- "input": [
1583
- "text"
1584
- ],
1585
- "output": [
1586
- "text"
1587
- ]
1588
- },
1589
- "open_weights": false,
1590
- "limit": {
1591
- "context": 8191,
1592
- "output": 1536
1593
- },
1594
- "cost": {
1595
- "input": 0.02,
1596
- "output": 0
1597
- }
1598
- },
1599
- "gpt-4o-mini": {
1600
- "id": "gpt-4o-mini",
1601
- "name": "GPT-4o mini",
1602
- "description": "Small omni GPT for cheap multimodal assistance and production-scale traffic",
1188
+ "gpt-5.4-mini": {
1189
+ "id": "gpt-5.4-mini",
1190
+ "name": "GPT-5.4 mini",
1191
+ "description": "Strong small GPT for coding subagents, quick tool use, and high-volume work",
1603
1192
  "family": "gpt-mini",
1604
1193
  "attachment": true,
1605
- "reasoning": false,
1194
+ "reasoning": true,
1195
+ "reasoning_options": [
1196
+ {
1197
+ "type": "effort",
1198
+ "values": [
1199
+ "none",
1200
+ "low",
1201
+ "medium",
1202
+ "high",
1203
+ "xhigh"
1204
+ ]
1205
+ }
1206
+ ],
1606
1207
  "tool_call": true,
1607
1208
  "structured_output": true,
1608
- "temperature": true,
1609
- "knowledge": "2023-09",
1610
- "release_date": "2024-07-18",
1611
- "last_updated": "2024-07-18",
1209
+ "temperature": false,
1210
+ "knowledge": "2025-08-31",
1211
+ "release_date": "2026-03-17",
1212
+ "last_updated": "2026-03-17",
1612
1213
  "modalities": {
1613
1214
  "input": [
1614
1215
  "text",
1615
- "image",
1616
- "pdf"
1216
+ "image"
1617
1217
  ],
1618
1218
  "output": [
1619
1219
  "text"
@@ -1621,27 +1221,43 @@
1621
1221
  },
1622
1222
  "open_weights": false,
1623
1223
  "limit": {
1624
- "context": 128000,
1625
- "output": 16384
1224
+ "context": 400000,
1225
+ "input": 272000,
1226
+ "output": 128000
1227
+ },
1228
+ "experimental": {
1229
+ "modes": {
1230
+ "fast": {
1231
+ "cost": {
1232
+ "input": 1.5,
1233
+ "output": 9,
1234
+ "cache_read": 0.15
1235
+ },
1236
+ "provider": {
1237
+ "body": {
1238
+ "service_tier": "priority"
1239
+ }
1240
+ }
1241
+ }
1242
+ }
1626
1243
  },
1627
1244
  "cost": {
1628
- "input": 0.15,
1629
- "output": 0.6,
1245
+ "input": 0.75,
1246
+ "output": 4.5,
1630
1247
  "cache_read": 0.075
1631
1248
  }
1632
1249
  },
1633
- "gpt-5.1": {
1634
- "id": "gpt-5.1",
1635
- "name": "GPT-5.1",
1636
- "description": "Sharper GPT-5 generation for coding, product work, and tool-assisted tasks",
1637
- "family": "gpt",
1250
+ "o1": {
1251
+ "id": "o1",
1252
+ "name": "o1",
1253
+ "description": "O-series reasoning model for hard analysis, math, coding, and planning",
1254
+ "family": "o",
1638
1255
  "attachment": true,
1639
1256
  "reasoning": true,
1640
1257
  "reasoning_options": [
1641
1258
  {
1642
1259
  "type": "effort",
1643
1260
  "values": [
1644
- "none",
1645
1261
  "low",
1646
1262
  "medium",
1647
1263
  "high"
@@ -1651,13 +1267,14 @@
1651
1267
  "tool_call": true,
1652
1268
  "structured_output": true,
1653
1269
  "temperature": false,
1654
- "knowledge": "2024-09-30",
1655
- "release_date": "2025-11-13",
1656
- "last_updated": "2025-11-13",
1270
+ "knowledge": "2023-09",
1271
+ "release_date": "2024-12-05",
1272
+ "last_updated": "2024-12-05",
1657
1273
  "modalities": {
1658
1274
  "input": [
1659
1275
  "text",
1660
- "image"
1276
+ "image",
1277
+ "pdf"
1661
1278
  ],
1662
1279
  "output": [
1663
1280
  "text"
@@ -1665,21 +1282,21 @@
1665
1282
  },
1666
1283
  "open_weights": false,
1667
1284
  "limit": {
1668
- "context": 400000,
1669
- "input": 272000,
1670
- "output": 128000
1285
+ "context": 200000,
1286
+ "output": 100000
1671
1287
  },
1288
+ "status": "deprecated",
1672
1289
  "cost": {
1673
- "input": 1.25,
1674
- "output": 10,
1675
- "cache_read": 0.125
1290
+ "input": 15,
1291
+ "output": 60,
1292
+ "cache_read": 7.5
1676
1293
  }
1677
1294
  },
1678
- "gpt-5.6": {
1679
- "id": "gpt-5.6",
1680
- "name": "GPT-5.6",
1681
- "description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
1682
- "family": "gpt-sol",
1295
+ "gpt-5.6-luna": {
1296
+ "id": "gpt-5.6-luna",
1297
+ "name": "GPT-5.6 Luna",
1298
+ "description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
1299
+ "family": "gpt-luna",
1683
1300
  "attachment": true,
1684
1301
  "reasoning": true,
1685
1302
  "reasoning_options": [
@@ -1721,10 +1338,10 @@
1721
1338
  "modes": {
1722
1339
  "fast": {
1723
1340
  "cost": {
1724
- "input": 10,
1725
- "output": 60,
1726
- "cache_read": 1,
1727
- "cache_write": 12.5
1341
+ "input": 0.4,
1342
+ "output": 2.4,
1343
+ "cache_read": 0.04,
1344
+ "cache_write": 0.5
1728
1345
  },
1729
1346
  "provider": {
1730
1347
  "body": {
@@ -1744,16 +1361,16 @@
1744
1361
  }
1745
1362
  },
1746
1363
  "cost": {
1747
- "input": 5,
1748
- "output": 30,
1749
- "cache_read": 0.5,
1750
- "cache_write": 6.25,
1364
+ "input": 0.2,
1365
+ "output": 1.2,
1366
+ "cache_read": 0.02,
1367
+ "cache_write": 0.25,
1751
1368
  "tiers": [
1752
1369
  {
1753
- "input": 10,
1754
- "output": 45,
1755
- "cache_read": 1,
1756
- "cache_write": 12.5,
1370
+ "input": 0.4,
1371
+ "output": 1.8,
1372
+ "cache_read": 0.04,
1373
+ "cache_write": 0.5,
1757
1374
  "tier": {
1758
1375
  "type": "context",
1759
1376
  "size": 272000
@@ -1761,110 +1378,38 @@
1761
1378
  }
1762
1379
  ],
1763
1380
  "context_over_200k": {
1764
- "input": 10,
1765
- "output": 45,
1766
- "cache_read": 1,
1767
- "cache_write": 12.5
1768
- }
1769
- }
1770
- },
1771
- "o3-deep-research": {
1772
- "id": "o3-deep-research",
1773
- "name": "o3-deep-research",
1774
- "description": "Research model for long-horizon investigation, synthesis, and analytical reports",
1775
- "family": "o",
1776
- "attachment": true,
1777
- "reasoning": true,
1778
- "reasoning_options": [
1779
- {
1780
- "type": "effort",
1781
- "values": [
1782
- "medium"
1783
- ]
1381
+ "input": 0.4,
1382
+ "output": 1.8,
1383
+ "cache_read": 0.04,
1384
+ "cache_write": 0.5
1784
1385
  }
1785
- ],
1786
- "tool_call": true,
1787
- "temperature": false,
1788
- "knowledge": "2024-05",
1789
- "release_date": "2024-06-26",
1790
- "last_updated": "2024-06-26",
1791
- "modalities": {
1792
- "input": [
1793
- "text",
1794
- "image"
1795
- ],
1796
- "output": [
1797
- "text"
1798
- ]
1799
- },
1800
- "open_weights": false,
1801
- "limit": {
1802
- "context": 200000,
1803
- "output": 100000
1804
- },
1805
- "cost": {
1806
- "input": 10,
1807
- "output": 40,
1808
- "cache_read": 2.5
1809
1386
  }
1810
1387
  },
1811
- "gpt-4o-2024-08-06": {
1812
- "id": "gpt-4o-2024-08-06",
1813
- "name": "GPT-4o (2024-08-06)",
1814
- "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
1388
+ "gpt-5.2": {
1389
+ "id": "gpt-5.2",
1390
+ "name": "GPT-5.2",
1391
+ "description": "Reliable GPT generation for broad coding, writing, and tool-assisted product work",
1815
1392
  "family": "gpt",
1816
1393
  "attachment": true,
1817
- "reasoning": false,
1818
- "tool_call": true,
1819
- "structured_output": true,
1820
- "temperature": true,
1821
- "knowledge": "2023-09",
1822
- "release_date": "2024-08-06",
1823
- "last_updated": "2024-08-06",
1824
- "modalities": {
1825
- "input": [
1826
- "text",
1827
- "image"
1828
- ],
1829
- "output": [
1830
- "text"
1831
- ]
1832
- },
1833
- "open_weights": false,
1834
- "limit": {
1835
- "context": 128000,
1836
- "output": 16384
1837
- },
1838
- "cost": {
1839
- "input": 2.5,
1840
- "output": 10,
1841
- "cache_read": 1.25
1842
- }
1843
- },
1844
- "gpt-5-nano": {
1845
- "id": "gpt-5-nano",
1846
- "name": "GPT-5 Nano",
1847
- "description": "Tiny GPT-5 lane for routing, extraction, classification, and bulk jobs",
1848
- "family": "gpt-nano",
1849
- "attachment": true,
1850
1394
  "reasoning": true,
1851
1395
  "reasoning_options": [
1852
1396
  {
1853
1397
  "type": "effort",
1854
1398
  "values": [
1855
- "minimal",
1399
+ "none",
1856
1400
  "low",
1857
1401
  "medium",
1858
- "high"
1402
+ "high",
1403
+ "xhigh"
1859
1404
  ]
1860
1405
  }
1861
1406
  ],
1862
1407
  "tool_call": true,
1863
1408
  "structured_output": true,
1864
1409
  "temperature": false,
1865
- "knowledge": "2024-05-30",
1866
- "release_date": "2025-08-07",
1867
- "last_updated": "2025-08-07",
1410
+ "knowledge": "2025-08-31",
1411
+ "release_date": "2025-12-11",
1412
+ "last_updated": "2025-12-11",
1868
1413
  "modalities": {
1869
1414
  "input": [
1870
1415
  "text",
@@ -1881,71 +1426,41 @@
1881
1426
  "output": 128000
1882
1427
  },
1883
1428
  "cost": {
1884
- "input": 0.05,
1885
- "output": 0.4,
1886
- "cache_read": 0.005
1429
+ "input": 1.75,
1430
+ "output": 14,
1431
+ "cache_read": 0.175
1887
1432
  }
1888
1433
  },
1889
- "o4-mini": {
1890
- "id": "o4-mini",
1891
- "name": "o4-mini",
1892
- "description": "Fast o-series model for compact reasoning, coding, and tool use",
1893
- "family": "o-mini",
1434
+ "gpt-5.3-codex": {
1435
+ "id": "gpt-5.3-codex",
1436
+ "name": "GPT-5.3 Codex",
1437
+ "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
1438
+ "family": "gpt-codex",
1894
1439
  "attachment": true,
1895
1440
  "reasoning": true,
1896
1441
  "reasoning_options": [
1897
1442
  {
1898
1443
  "type": "effort",
1899
1444
  "values": [
1445
+ "none",
1900
1446
  "low",
1901
1447
  "medium",
1902
- "high"
1448
+ "high",
1449
+ "xhigh"
1903
1450
  ]
1904
1451
  }
1905
1452
  ],
1906
1453
  "tool_call": true,
1907
1454
  "structured_output": true,
1908
1455
  "temperature": false,
1909
- "knowledge": "2024-05",
1910
- "release_date": "2025-04-16",
1911
- "last_updated": "2025-04-16",
1912
- "modalities": {
1913
- "input": [
1914
- "text",
1915
- "image"
1916
- ],
1917
- "output": [
1918
- "text"
1919
- ]
1920
- },
1921
- "open_weights": false,
1922
- "limit": {
1923
- "context": 200000,
1924
- "output": 100000
1925
- },
1926
- "cost": {
1927
- "input": 1.1,
1928
- "output": 4.4,
1929
- "cache_read": 0.275
1930
- }
1931
- },
1932
- "gpt-4-turbo": {
1933
- "id": "gpt-4-turbo",
1934
- "name": "GPT-4 Turbo",
1935
- "description": "Compact GPT model for low-latency assistance and high-volume workloads",
1936
- "family": "gpt",
1937
- "attachment": true,
1938
- "reasoning": false,
1939
- "tool_call": true,
1940
- "structured_output": false,
1941
- "temperature": true,
1942
- "knowledge": "2023-12",
1943
- "release_date": "2023-11-06",
1944
- "last_updated": "2024-04-09",
1456
+ "knowledge": "2025-08-31",
1457
+ "release_date": "2026-02-05",
1458
+ "last_updated": "2026-02-05",
1945
1459
  "modalities": {
1946
1460
  "input": [
1947
1461
  "text",
1948
- "image"
1462
+ "image",
1463
+ "pdf"
1949
1464
  ],
1950
1465
  "output": [
1951
1466
  "text"
@@ -1953,39 +1468,40 @@
1953
1468
  },
1954
1469
  "open_weights": false,
1955
1470
  "limit": {
1956
- "context": 128000,
1957
- "output": 4096
1471
+ "context": 400000,
1472
+ "input": 272000,
1473
+ "output": 128000
1958
1474
  },
1959
1475
  "cost": {
1960
- "input": 10,
1961
- "output": 30
1476
+ "input": 1.75,
1477
+ "output": 14,
1478
+ "cache_read": 0.175
1962
1479
  }
1963
1480
  },
1964
- "gpt-5.2": {
1965
- "id": "gpt-5.2",
1966
- "name": "GPT-5.2",
1967
- "description": "Reliable GPT generation for broad coding, writing, and tool-assisted product work",
1968
- "family": "gpt",
1481
+ "gpt-5-mini": {
1482
+ "id": "gpt-5-mini",
1483
+ "name": "GPT-5 Mini",
1484
+ "description": "Small GPT-5 for responsive agents, coding help, and everyday automation",
1485
+ "family": "gpt-mini",
1969
1486
  "attachment": true,
1970
1487
  "reasoning": true,
1971
1488
  "reasoning_options": [
1972
1489
  {
1973
1490
  "type": "effort",
1974
1491
  "values": [
1975
- "none",
1492
+ "minimal",
1976
1493
  "low",
1977
1494
  "medium",
1978
- "high",
1979
- "xhigh"
1495
+ "high"
1980
1496
  ]
1981
1497
  }
1982
1498
  ],
1983
1499
  "tool_call": true,
1984
1500
  "structured_output": true,
1985
1501
  "temperature": false,
1986
- "knowledge": "2025-08-31",
1987
- "release_date": "2025-12-11",
1988
- "last_updated": "2025-12-11",
1502
+ "knowledge": "2024-05-30",
1503
+ "release_date": "2025-08-07",
1504
+ "last_updated": "2025-08-07",
1989
1505
  "modalities": {
1990
1506
  "input": [
1991
1507
  "text",
@@ -2002,29 +1518,38 @@
2002
1518
  "output": 128000
2003
1519
  },
2004
1520
  "cost": {
2005
- "input": 1.75,
2006
- "output": 14,
2007
- "cache_read": 0.175
1521
+ "input": 0.25,
1522
+ "output": 2,
1523
+ "cache_read": 0.025
2008
1524
  }
2009
1525
  },
2010
- "gpt-4o": {
2011
- "id": "gpt-4o",
2012
- "name": "GPT-4o",
2013
- "description": "Omni-era GPT for multimodal chat, practical coding, and general assistants",
2014
- "family": "gpt",
1526
+ "o1-pro": {
1527
+ "id": "o1-pro",
1528
+ "name": "o1-pro",
1529
+ "description": "O-series reasoning model for hard analysis, math, coding, and planning",
1530
+ "family": "o-pro",
2015
1531
  "attachment": true,
2016
- "reasoning": false,
1532
+ "reasoning": true,
1533
+ "reasoning_options": [
1534
+ {
1535
+ "type": "effort",
1536
+ "values": [
1537
+ "low",
1538
+ "medium",
1539
+ "high"
1540
+ ]
1541
+ }
1542
+ ],
2017
1543
  "tool_call": true,
2018
1544
  "structured_output": true,
2019
- "temperature": true,
1545
+ "temperature": false,
2020
1546
  "knowledge": "2023-09",
2021
- "release_date": "2024-05-13",
2022
- "last_updated": "2024-08-06",
1547
+ "release_date": "2025-03-19",
1548
+ "last_updated": "2025-03-19",
2023
1549
  "modalities": {
2024
1550
  "input": [
2025
1551
  "text",
2026
- "image",
2027
- "pdf"
1552
+ "image"
2028
1553
  ],
2029
1554
  "output": [
2030
1555
  "text"
@@ -2032,20 +1557,20 @@
2032
1557
  },
2033
1558
  "open_weights": false,
2034
1559
  "limit": {
2035
- "context": 128000,
2036
- "output": 16384
1560
+ "context": 200000,
1561
+ "output": 100000
2037
1562
  },
1563
+ "status": "deprecated",
2038
1564
  "cost": {
2039
- "input": 2.5,
2040
- "output": 10,
2041
- "cache_read": 1.25
1565
+ "input": 150,
1566
+ "output": 600
2042
1567
  }
2043
1568
  },
2044
- "gpt-5.6-luna": {
2045
- "id": "gpt-5.6-luna",
2046
- "name": "GPT-5.6 Luna",
2047
- "description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
2048
- "family": "gpt-luna",
1569
+ "gpt-5.6": {
1570
+ "id": "gpt-5.6",
1571
+ "name": "GPT-5.6",
1572
+ "description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
1573
+ "family": "gpt-sol",
2049
1574
  "attachment": true,
2050
1575
  "reasoning": true,
2051
1576
  "reasoning_options": [
@@ -2087,10 +1612,10 @@
2087
1612
  "modes": {
2088
1613
  "fast": {
2089
1614
  "cost": {
2090
- "input": 2,
2091
- "output": 12,
2092
- "cache_read": 0.2,
2093
- "cache_write": 2.5
1615
+ "input": 10,
1616
+ "output": 60,
1617
+ "cache_read": 1,
1618
+ "cache_write": 12.5
2094
1619
  },
2095
1620
  "provider": {
2096
1621
  "body": {
@@ -2110,16 +1635,16 @@
2110
1635
  }
2111
1636
  },
2112
1637
  "cost": {
2113
- "input": 1,
2114
- "output": 6,
2115
- "cache_read": 0.1,
2116
- "cache_write": 1.25,
1638
+ "input": 5,
1639
+ "output": 30,
1640
+ "cache_read": 0.5,
1641
+ "cache_write": 6.25,
2117
1642
  "tiers": [
2118
1643
  {
2119
- "input": 2,
2120
- "output": 9,
2121
- "cache_read": 0.2,
2122
- "cache_write": 2.5,
1644
+ "input": 10,
1645
+ "output": 45,
1646
+ "cache_read": 1,
1647
+ "cache_write": 12.5,
2123
1648
  "tier": {
2124
1649
  "type": "context",
2125
1650
  "size": 272000
@@ -2127,79 +1652,138 @@
2127
1652
  }
2128
1653
  ],
2129
1654
  "context_over_200k": {
2130
- "input": 2,
2131
- "output": 9,
2132
- "cache_read": 0.2,
2133
- "cache_write": 2.5
1655
+ "input": 10,
1656
+ "output": 45,
1657
+ "cache_read": 1,
1658
+ "cache_write": 12.5
2134
1659
  }
2135
1660
  }
2136
1661
  },
2137
- "chatgpt-image-latest": {
2138
- "id": "chatgpt-image-latest",
2139
- "name": "chatgpt-image-latest",
2140
- "description": "Image model for prompt-driven generation, editing, and visual design workflows",
2141
- "family": "gpt-image",
1662
+ "gpt-5.1": {
1663
+ "id": "gpt-5.1",
1664
+ "name": "GPT-5.1",
1665
+ "description": "Sharper GPT-5 generation for coding, product work, and tool-assisted tasks",
1666
+ "family": "gpt",
2142
1667
  "attachment": true,
2143
- "reasoning": false,
2144
- "tool_call": false,
1668
+ "reasoning": true,
1669
+ "reasoning_options": [
1670
+ {
1671
+ "type": "effort",
1672
+ "values": [
1673
+ "none",
1674
+ "low",
1675
+ "medium",
1676
+ "high"
1677
+ ]
1678
+ }
1679
+ ],
1680
+ "tool_call": true,
1681
+ "structured_output": true,
2145
1682
  "temperature": false,
2146
- "release_date": "2025-12-16",
2147
- "last_updated": "2025-12-16",
1683
+ "knowledge": "2024-09-30",
1684
+ "release_date": "2025-11-13",
1685
+ "last_updated": "2025-11-13",
2148
1686
  "modalities": {
2149
1687
  "input": [
2150
1688
  "text",
2151
1689
  "image"
2152
1690
  ],
2153
1691
  "output": [
1692
+ "text"
1693
+ ]
1694
+ },
1695
+ "open_weights": false,
1696
+ "limit": {
1697
+ "context": 400000,
1698
+ "input": 272000,
1699
+ "output": 128000
1700
+ },
1701
+ "cost": {
1702
+ "input": 1.25,
1703
+ "output": 10,
1704
+ "cache_read": 0.125
1705
+ }
1706
+ },
1707
+ "gpt-4-turbo": {
1708
+ "id": "gpt-4-turbo",
1709
+ "name": "GPT-4 Turbo",
1710
+ "description": "Compact GPT model for low-latency assistance and high-volume workloads",
1711
+ "family": "gpt",
1712
+ "attachment": true,
1713
+ "reasoning": false,
1714
+ "tool_call": true,
1715
+ "structured_output": false,
1716
+ "temperature": true,
1717
+ "knowledge": "2023-12",
1718
+ "release_date": "2023-11-06",
1719
+ "last_updated": "2024-04-09",
1720
+ "modalities": {
1721
+ "input": [
2154
1722
  "text",
2155
1723
  "image"
1724
+ ],
1725
+ "output": [
1726
+ "text"
2156
1727
  ]
2157
1728
  },
2158
1729
  "open_weights": false,
2159
1730
  "limit": {
2160
- "context": 0,
2161
- "input": 0,
2162
- "output": 0
1731
+ "context": 128000,
1732
+ "output": 4096
1733
+ },
1734
+ "status": "deprecated",
1735
+ "cost": {
1736
+ "input": 10,
1737
+ "output": 30
2163
1738
  }
2164
1739
  },
2165
- "gpt-image-1": {
2166
- "id": "gpt-image-1",
2167
- "name": "gpt-image-1",
2168
- "description": "OpenAI image model for production generation, edits, and brand-safe visual workflows",
2169
- "family": "gpt-image",
1740
+ "gpt-4o-2024-08-06": {
1741
+ "id": "gpt-4o-2024-08-06",
1742
+ "name": "GPT-4o (2024-08-06)",
1743
+ "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
1744
+ "family": "gpt",
2170
1745
  "attachment": true,
2171
1746
  "reasoning": false,
2172
- "tool_call": false,
2173
- "temperature": false,
2174
- "release_date": "2025-04-24",
2175
- "last_updated": "2025-04-24",
1747
+ "tool_call": true,
1748
+ "structured_output": true,
1749
+ "temperature": true,
1750
+ "knowledge": "2023-09",
1751
+ "release_date": "2024-08-06",
1752
+ "last_updated": "2024-08-06",
2176
1753
  "modalities": {
2177
1754
  "input": [
2178
1755
  "text",
2179
1756
  "image"
2180
1757
  ],
2181
1758
  "output": [
2182
- "image"
1759
+ "text"
2183
1760
  ]
2184
1761
  },
2185
1762
  "open_weights": false,
2186
1763
  "limit": {
2187
- "context": 0,
2188
- "input": 0,
2189
- "output": 0
1764
+ "context": 128000,
1765
+ "output": 16384
1766
+ },
1767
+ "cost": {
1768
+ "input": 2.5,
1769
+ "output": 10,
1770
+ "cache_read": 1.25
2190
1771
  }
2191
1772
  },
2192
- "gpt-5-pro": {
2193
- "id": "gpt-5-pro",
2194
- "name": "GPT-5 Pro",
2195
- "description": "Higher-accuracy GPT-5 tier for tough analysis, coding reviews, and planning",
2196
- "family": "gpt-pro",
1773
+ "gpt-5-nano": {
1774
+ "id": "gpt-5-nano",
1775
+ "name": "GPT-5 Nano",
1776
+ "description": "Tiny GPT-5 lane for routing, extraction, classification, and bulk jobs",
1777
+ "family": "gpt-nano",
2197
1778
  "attachment": true,
2198
1779
  "reasoning": true,
2199
1780
  "reasoning_options": [
2200
1781
  {
2201
1782
  "type": "effort",
2202
1783
  "values": [
1784
+ "minimal",
1785
+ "low",
1786
+ "medium",
2203
1787
  "high"
2204
1788
  ]
2205
1789
  }
@@ -2207,9 +1791,9 @@
2207
1791
  "tool_call": true,
2208
1792
  "structured_output": true,
2209
1793
  "temperature": false,
2210
- "knowledge": "2024-09-30",
2211
- "release_date": "2025-10-06",
2212
- "last_updated": "2025-10-06",
1794
+ "knowledge": "2024-05-30",
1795
+ "release_date": "2025-08-07",
1796
+ "last_updated": "2025-08-07",
2213
1797
  "modalities": {
2214
1798
  "input": [
2215
1799
  "text",
@@ -2223,40 +1807,42 @@
2223
1807
  "limit": {
2224
1808
  "context": 400000,
2225
1809
  "input": 272000,
2226
- "output": 272000
1810
+ "output": 128000
2227
1811
  },
2228
1812
  "cost": {
2229
- "input": 15,
2230
- "output": 120
1813
+ "input": 0.05,
1814
+ "output": 0.4,
1815
+ "cache_read": 0.005
2231
1816
  }
2232
1817
  },
2233
- "gpt-5.2-pro": {
2234
- "id": "gpt-5.2-pro",
2235
- "name": "GPT-5.2 Pro",
2236
- "description": "Higher-accuracy GPT-5.2 variant for tougher reasoning and review workflows",
2237
- "family": "gpt-pro",
1818
+ "o3": {
1819
+ "id": "o3",
1820
+ "name": "o3",
1821
+ "description": "Deliberate o-series reasoner for hard math, coding, and multi-step analysis",
1822
+ "family": "o",
2238
1823
  "attachment": true,
2239
1824
  "reasoning": true,
2240
1825
  "reasoning_options": [
2241
1826
  {
2242
1827
  "type": "effort",
2243
1828
  "values": [
1829
+ "low",
2244
1830
  "medium",
2245
- "high",
2246
- "xhigh"
1831
+ "high"
2247
1832
  ]
2248
1833
  }
2249
1834
  ],
2250
1835
  "tool_call": true,
2251
- "structured_output": false,
1836
+ "structured_output": true,
2252
1837
  "temperature": false,
2253
- "knowledge": "2025-08-31",
2254
- "release_date": "2025-12-11",
2255
- "last_updated": "2025-12-11",
1838
+ "knowledge": "2024-05",
1839
+ "release_date": "2025-04-16",
1840
+ "last_updated": "2025-04-16",
2256
1841
  "modalities": {
2257
1842
  "input": [
2258
1843
  "text",
2259
- "image"
1844
+ "image",
1845
+ "pdf"
2260
1846
  ],
2261
1847
  "output": [
2262
1848
  "text"
@@ -2264,20 +1850,20 @@
2264
1850
  },
2265
1851
  "open_weights": false,
2266
1852
  "limit": {
2267
- "context": 400000,
2268
- "input": 272000,
2269
- "output": 128000
1853
+ "context": 200000,
1854
+ "output": 100000
2270
1855
  },
2271
1856
  "cost": {
2272
- "input": 21,
2273
- "output": 168
1857
+ "input": 2,
1858
+ "output": 8,
1859
+ "cache_read": 0.5
2274
1860
  }
2275
1861
  },
2276
- "gpt-5.6-sol": {
2277
- "id": "gpt-5.6-sol",
2278
- "name": "GPT-5.6 Sol",
2279
- "description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
2280
- "family": "gpt-sol",
1862
+ "gpt-5.6-terra": {
1863
+ "id": "gpt-5.6-terra",
1864
+ "name": "GPT-5.6 Terra",
1865
+ "description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
1866
+ "family": "gpt-terra",
2281
1867
  "attachment": true,
2282
1868
  "reasoning": true,
2283
1869
  "reasoning_options": [
@@ -2319,10 +1905,10 @@
2319
1905
  "modes": {
2320
1906
  "fast": {
2321
1907
  "cost": {
2322
- "input": 10,
2323
- "output": 60,
2324
- "cache_read": 1,
2325
- "cache_write": 12.5
1908
+ "input": 4,
1909
+ "output": 24,
1910
+ "cache_read": 0.4,
1911
+ "cache_write": 5
2326
1912
  },
2327
1913
  "provider": {
2328
1914
  "body": {
@@ -2342,16 +1928,16 @@
2342
1928
  }
2343
1929
  },
2344
1930
  "cost": {
2345
- "input": 5,
2346
- "output": 30,
2347
- "cache_read": 0.5,
2348
- "cache_write": 6.25,
1931
+ "input": 2,
1932
+ "output": 12,
1933
+ "cache_read": 0.2,
1934
+ "cache_write": 2.5,
2349
1935
  "tiers": [
2350
1936
  {
2351
- "input": 10,
2352
- "output": 45,
2353
- "cache_read": 1,
2354
- "cache_write": 12.5,
1937
+ "input": 4,
1938
+ "output": 18,
1939
+ "cache_read": 0.4,
1940
+ "cache_write": 5,
2355
1941
  "tier": {
2356
1942
  "type": "context",
2357
1943
  "size": 272000
@@ -2359,18 +1945,80 @@
2359
1945
  }
2360
1946
  ],
2361
1947
  "context_over_200k": {
2362
- "input": 10,
2363
- "output": 45,
2364
- "cache_read": 1,
2365
- "cache_write": 12.5
1948
+ "input": 4,
1949
+ "output": 18,
1950
+ "cache_read": 0.4,
1951
+ "cache_write": 5
2366
1952
  }
2367
1953
  }
2368
1954
  },
2369
- "o3-pro": {
2370
- "id": "o3-pro",
2371
- "name": "o3-pro",
2372
- "description": "High-effort o3 tier for difficult technical reasoning and careful answers",
2373
- "family": "o-pro",
1955
+ "gpt-image-1": {
1956
+ "id": "gpt-image-1",
1957
+ "name": "gpt-image-1",
1958
+ "description": "OpenAI image model for production generation, edits, and brand-safe visual workflows",
1959
+ "family": "gpt-image",
1960
+ "attachment": true,
1961
+ "reasoning": false,
1962
+ "tool_call": false,
1963
+ "temperature": false,
1964
+ "release_date": "2025-04-24",
1965
+ "last_updated": "2025-04-24",
1966
+ "modalities": {
1967
+ "input": [
1968
+ "text",
1969
+ "image"
1970
+ ],
1971
+ "output": [
1972
+ "image"
1973
+ ]
1974
+ },
1975
+ "open_weights": false,
1976
+ "limit": {
1977
+ "context": 0,
1978
+ "input": 0,
1979
+ "output": 0
1980
+ },
1981
+ "status": "deprecated"
1982
+ },
1983
+ "gpt-4.1": {
1984
+ "id": "gpt-4.1",
1985
+ "name": "GPT-4.1",
1986
+ "description": "Long-lived GPT workhorse for coding, instruction following, and production apps",
1987
+ "family": "gpt",
1988
+ "attachment": true,
1989
+ "reasoning": false,
1990
+ "tool_call": true,
1991
+ "structured_output": true,
1992
+ "temperature": true,
1993
+ "knowledge": "2024-04",
1994
+ "release_date": "2025-04-14",
1995
+ "last_updated": "2025-04-14",
1996
+ "modalities": {
1997
+ "input": [
1998
+ "text",
1999
+ "image",
2000
+ "pdf"
2001
+ ],
2002
+ "output": [
2003
+ "text"
2004
+ ]
2005
+ },
2006
+ "open_weights": false,
2007
+ "limit": {
2008
+ "context": 1047576,
2009
+ "output": 32768
2010
+ },
2011
+ "cost": {
2012
+ "input": 2,
2013
+ "output": 8,
2014
+ "cache_read": 0.5
2015
+ }
2016
+ },
2017
+ "o4-mini": {
2018
+ "id": "o4-mini",
2019
+ "name": "o4-mini",
2020
+ "description": "Fast o-series model for compact reasoning, coding, and tool use",
2021
+ "family": "o-mini",
2374
2022
  "attachment": true,
2375
2023
  "reasoning": true,
2376
2024
  "reasoning_options": [
@@ -2387,8 +2035,8 @@
2387
2035
  "structured_output": true,
2388
2036
  "temperature": false,
2389
2037
  "knowledge": "2024-05",
2390
- "release_date": "2025-06-10",
2391
- "last_updated": "2025-06-10",
2038
+ "release_date": "2025-04-16",
2039
+ "last_updated": "2025-04-16",
2392
2040
  "modalities": {
2393
2041
  "input": [
2394
2042
  "text",
@@ -2403,41 +2051,29 @@
2403
2051
  "context": 200000,
2404
2052
  "output": 100000
2405
2053
  },
2054
+ "status": "deprecated",
2406
2055
  "cost": {
2407
- "input": 20,
2408
- "output": 80
2056
+ "input": 1.1,
2057
+ "output": 4.4,
2058
+ "cache_read": 0.275
2409
2059
  }
2410
2060
  },
2411
- "gpt-5.5": {
2412
- "id": "gpt-5.5",
2413
- "name": "GPT-5.5",
2414
- "description": "Default frontier GPT for coding, computer use, research, and knowledge work",
2061
+ "gpt-4": {
2062
+ "id": "gpt-4",
2063
+ "name": "GPT-4",
2064
+ "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
2415
2065
  "family": "gpt",
2416
2066
  "attachment": true,
2417
- "reasoning": true,
2418
- "reasoning_options": [
2419
- {
2420
- "type": "effort",
2421
- "values": [
2422
- "none",
2423
- "low",
2424
- "medium",
2425
- "high",
2426
- "xhigh"
2427
- ]
2428
- }
2429
- ],
2067
+ "reasoning": false,
2430
2068
  "tool_call": true,
2431
- "structured_output": true,
2432
- "temperature": false,
2433
- "knowledge": "2025-12-01",
2434
- "release_date": "2026-04-23",
2435
- "last_updated": "2026-04-23",
2069
+ "structured_output": false,
2070
+ "temperature": true,
2071
+ "knowledge": "2023-11",
2072
+ "release_date": "2023-11-06",
2073
+ "last_updated": "2024-04-09",
2436
2074
  "modalities": {
2437
2075
  "input": [
2438
- "text",
2439
- "image",
2440
- "pdf"
2076
+ "text"
2441
2077
  ],
2442
2078
  "output": [
2443
2079
  "text"
@@ -2445,78 +2081,73 @@
2445
2081
  },
2446
2082
  "open_weights": false,
2447
2083
  "limit": {
2448
- "context": 1050000,
2449
- "input": 922000,
2450
- "output": 128000
2451
- },
2452
- "experimental": {
2453
- "modes": {
2454
- "fast": {
2455
- "cost": {
2456
- "input": 12.5,
2457
- "output": 75,
2458
- "cache_read": 1.25
2459
- },
2460
- "provider": {
2461
- "body": {
2462
- "service_tier": "priority"
2463
- }
2464
- }
2465
- }
2466
- }
2084
+ "context": 8192,
2085
+ "output": 8192
2467
2086
  },
2087
+ "status": "deprecated",
2468
2088
  "cost": {
2469
- "input": 5,
2470
- "output": 30,
2471
- "cache_read": 0.5,
2472
- "tiers": [
2473
- {
2474
- "input": 10,
2475
- "output": 45,
2476
- "cache_read": 1,
2477
- "tier": {
2478
- "type": "context",
2479
- "size": 272000
2480
- }
2481
- }
2482
- ],
2483
- "context_over_200k": {
2484
- "input": 10,
2485
- "output": 45,
2486
- "cache_read": 1
2487
- }
2089
+ "input": 30,
2090
+ "output": 60
2488
2091
  }
2489
2092
  },
2490
- "gpt-image-2": {
2491
- "id": "gpt-image-2",
2492
- "name": "gpt-image-2",
2493
- "description": "Image model for prompt-driven generation, editing, and visual design workflows",
2494
- "family": "gpt-image",
2495
- "attachment": true,
2093
+ "text-embedding-3-large": {
2094
+ "id": "text-embedding-3-large",
2095
+ "name": "text-embedding-3-large",
2096
+ "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
2097
+ "family": "text-embedding",
2098
+ "attachment": false,
2496
2099
  "reasoning": false,
2497
2100
  "tool_call": false,
2498
2101
  "temperature": false,
2499
- "release_date": "2026-04-21",
2500
- "last_updated": "2026-04-21",
2102
+ "knowledge": "2024-01",
2103
+ "release_date": "2024-01-25",
2104
+ "last_updated": "2024-01-25",
2501
2105
  "modalities": {
2502
2106
  "input": [
2503
- "text",
2504
- "image"
2107
+ "text"
2505
2108
  ],
2506
2109
  "output": [
2507
- "image"
2110
+ "text"
2508
2111
  ]
2509
2112
  },
2510
2113
  "open_weights": false,
2511
2114
  "limit": {
2512
- "context": 0,
2513
- "input": 0,
2115
+ "context": 8191,
2116
+ "output": 3072
2117
+ },
2118
+ "cost": {
2119
+ "input": 0.13,
2514
2120
  "output": 0
2121
+ }
2122
+ },
2123
+ "text-embedding-3-small": {
2124
+ "id": "text-embedding-3-small",
2125
+ "name": "text-embedding-3-small",
2126
+ "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
2127
+ "family": "text-embedding",
2128
+ "attachment": false,
2129
+ "reasoning": false,
2130
+ "tool_call": false,
2131
+ "temperature": false,
2132
+ "knowledge": "2024-01",
2133
+ "release_date": "2024-01-25",
2134
+ "last_updated": "2024-01-25",
2135
+ "modalities": {
2136
+ "input": [
2137
+ "text"
2138
+ ],
2139
+ "output": [
2140
+ "text"
2141
+ ]
2142
+ },
2143
+ "open_weights": false,
2144
+ "limit": {
2145
+ "context": 8191,
2146
+ "output": 1536
2515
2147
  },
2516
2148
  "cost": {
2517
- "input": 5,
2518
- "output": 30,
2519
- "cache_read": 1.25
2149
+ "input": 0.02,
2150
+ "output": 0
2520
2151
  }
2521
2152
  }
2522
2153
  }