llm.rb 15.0.3 → 15.2.0

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 (104) hide show
  1. checksums.yaml +4 -4
  2. data/CHANGELOG.md +433 -3
  3. data/README.md +188 -71
  4. data/bin/llm.rb +50 -7
  5. data/data/alibaba.json +912 -823
  6. data/data/anthropic.json +234 -187
  7. data/data/bedrock.json +3702 -2058
  8. data/data/deepinfra.json +1288 -951
  9. data/data/deepseek.json +87 -53
  10. data/data/google.json +670 -670
  11. data/data/mistral.json +501 -460
  12. data/data/moonshot.json +43 -248
  13. data/data/openai.json +1008 -914
  14. data/data/openrouter.json +14417 -0
  15. data/data/xai.json +213 -201
  16. data/data/zai.json +242 -149
  17. data/docs/deepdive/advanced/compaction.md +5 -5
  18. data/docs/deepdive/advanced/context.md +8 -6
  19. data/docs/deepdive/advanced/guard.md +2 -2
  20. data/docs/deepdive/features/builtin_tools.md +93 -22
  21. data/docs/deepdive/features/{repl.md → console.md} +28 -28
  22. data/docs/deepdive/features/database.md +3 -3
  23. data/docs/deepdive/fundamentals/agents.md +13 -12
  24. data/docs/deepdive/fundamentals/providers.md +91 -6
  25. data/docs/deepdive/fundamentals/skills.md +14 -6
  26. data/docs/deepdive/fundamentals/stream.md +4 -4
  27. data/docs/deepdive/fundamentals/tools.md +63 -31
  28. data/docs/deepdive/reference/cost.md +2 -2
  29. data/docs/deepdive/reference/model_registry.md +2 -2
  30. data/docs/deepdive/reference/tracer.md +15 -13
  31. data/docs/deepdive.md +2 -2
  32. data/lib/llm/active_record/acts_as_agent.rb +9 -5
  33. data/lib/llm/agent.rb +40 -15
  34. data/lib/llm/{repl → console}/bar.rb +3 -3
  35. data/lib/llm/{repl → console}/buffer.rb +24 -9
  36. data/lib/llm/{repl → console}/color.rb +2 -2
  37. data/lib/llm/{repl → console}/command.rb +12 -12
  38. data/lib/llm/{repl → console}/commands/exit.rb +4 -4
  39. data/lib/llm/{repl → console}/commands/help.rb +1 -1
  40. data/lib/llm/{repl/commands/compact.rb → console/commands/keep.rb} +11 -9
  41. data/lib/llm/{repl → console}/commands/model.rb +2 -2
  42. data/lib/llm/{repl → console}/input/cache.rb +2 -2
  43. data/lib/llm/{repl → console}/input/char.rb +2 -2
  44. data/lib/llm/{repl → console}/input/row.rb +1 -1
  45. data/lib/llm/{repl → console}/input.rb +18 -10
  46. data/lib/llm/console/markdown/parser.rb +78 -0
  47. data/lib/llm/{repl → console}/markdown/table.rb +8 -5
  48. data/lib/llm/{repl → console}/markdown.rb +13 -30
  49. data/lib/llm/console/node.rb +69 -0
  50. data/lib/llm/{repl → console}/status.rb +11 -11
  51. data/lib/llm/{repl → console}/stream.rb +36 -9
  52. data/lib/llm/{repl → console}/walker.rb +1 -1
  53. data/lib/llm/{repl → console}/window.rb +17 -17
  54. data/lib/llm/{repl.rb → console.rb} +39 -19
  55. data/lib/llm/context/deserializer.rb +2 -1
  56. data/lib/llm/context.rb +29 -13
  57. data/lib/llm/cost.rb +13 -0
  58. data/lib/llm/function/async/reactor.rb +20 -1
  59. data/lib/llm/function/fork/task.rb +14 -10
  60. data/lib/llm/function.rb +1 -1
  61. data/lib/llm/json_adapter.rb +40 -28
  62. data/lib/llm/message.rb +7 -0
  63. data/lib/llm/provider.rb +31 -10
  64. data/lib/llm/providers/alibaba.rb +1 -1
  65. data/lib/llm/providers/anthropic.rb +1 -1
  66. data/lib/llm/providers/bedrock/models.rb +2 -2
  67. data/lib/llm/providers/bedrock.rb +1 -1
  68. data/lib/llm/providers/deepseek.rb +1 -1
  69. data/lib/llm/providers/google.rb +1 -1
  70. data/lib/llm/providers/ollama.rb +1 -1
  71. data/lib/llm/providers/openai/responses.rb +2 -1
  72. data/lib/llm/providers/openai.rb +2 -1
  73. data/lib/llm/providers/openrouter.rb +87 -0
  74. data/lib/llm/schema/leaf.rb +34 -2
  75. data/lib/llm/schema.rb +4 -2
  76. data/lib/llm/sequel/agent.rb +9 -5
  77. data/lib/llm/skill.rb +7 -1
  78. data/lib/llm/stream.rb +8 -3
  79. data/lib/llm/tool/param.rb +5 -1
  80. data/lib/llm/tool.rb +5 -0
  81. data/lib/llm/tools/bundle.rb +53 -0
  82. data/lib/llm/tools/edit-file.rb +7 -2
  83. data/lib/llm/tools/exec.rb +78 -0
  84. data/lib/llm/tools/git.rb +27 -26
  85. data/lib/llm/tools/mkdir.rb +12 -19
  86. data/lib/llm/tools/read_file.rb +69 -9
  87. data/lib/llm/tools/rg.rb +20 -24
  88. data/lib/llm/tools/ruby.rb +17 -25
  89. data/lib/llm/tools/utils.rb +75 -2
  90. data/lib/llm/tools/write_file.rb +4 -1
  91. data/lib/llm/tracer/logger.rb +2 -2
  92. data/lib/llm/tracer/pretty_logger.rb +4 -4
  93. data/lib/llm/tracer/telemetry.rb +2 -2
  94. data/lib/llm/tracer.rb +33 -0
  95. data/lib/llm/transport/curb.rb +5 -3
  96. data/lib/llm/transport/http.rb +5 -2
  97. data/lib/llm/transport/persistent_http.rb +6 -4
  98. data/lib/llm/transport/utils.rb +8 -6
  99. data/lib/llm/version.rb +1 -1
  100. data/lib/llm.rb +18 -12
  101. data/llm.gemspec +8 -8
  102. metadata +80 -37
  103. data/lib/llm/repl/node.rb +0 -44
  104. data/lib/llm/tools/shell.rb +0 -55
data/data/google.json CHANGED
@@ -9,37 +9,6 @@
9
9
  "name": "Google",
10
10
  "doc": "https://ai.google.dev/gemini-api/docs/models",
11
11
  "models": {
12
- "gemini-3.1-flash-tts-preview": {
13
- "id": "gemini-3.1-flash-tts-preview",
14
- "name": "Gemini 3.1 Flash TTS Preview",
15
- "description": "Low-latency speech generation with steerable prompts and expressive audio tags",
16
- "family": "gemini-flash",
17
- "attachment": false,
18
- "reasoning": true,
19
- "reasoning_options": [],
20
- "tool_call": false,
21
- "temperature": true,
22
- "knowledge": "2025-01",
23
- "release_date": "2026-04-15",
24
- "last_updated": "2026-04-15",
25
- "modalities": {
26
- "input": [
27
- "text"
28
- ],
29
- "output": [
30
- "audio"
31
- ]
32
- },
33
- "open_weights": false,
34
- "limit": {
35
- "context": 8192,
36
- "output": 16384
37
- },
38
- "cost": {
39
- "input": 1,
40
- "output": 20
41
- }
42
- },
43
12
  "gemma-4-26b-a4b-it": {
44
13
  "id": "gemma-4-26b-a4b-it",
45
14
  "name": "Gemma 4 26B A4B IT",
@@ -72,10 +41,10 @@
72
41
  "output": 32768
73
42
  }
74
43
  },
75
- "gemini-3-pro-image": {
76
- "id": "gemini-3-pro-image",
77
- "name": "Nano Banana Pro",
78
- "description": "Nano Banana Pro for higher-fidelity image generation and design-heavy edits",
44
+ "gemini-3.1-pro-preview-customtools": {
45
+ "id": "gemini-3.1-pro-preview-customtools",
46
+ "name": "Gemini 3.1 Pro Preview Custom Tools",
47
+ "description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
79
48
  "family": "gemini-pro",
80
49
  "attachment": true,
81
50
  "reasoning": true,
@@ -84,72 +53,60 @@
84
53
  "type": "effort",
85
54
  "values": [
86
55
  "low",
56
+ "medium",
87
57
  "high"
88
58
  ]
89
59
  }
90
60
  ],
91
- "tool_call": false,
61
+ "tool_call": true,
62
+ "structured_output": true,
92
63
  "temperature": true,
93
64
  "knowledge": "2025-01",
94
- "release_date": "2026-05-28",
95
- "last_updated": "2026-05-28",
65
+ "release_date": "2026-02-19",
66
+ "last_updated": "2026-02-19",
96
67
  "modalities": {
97
68
  "input": [
98
69
  "text",
99
- "image"
70
+ "image",
71
+ "video",
72
+ "audio",
73
+ "pdf"
100
74
  ],
101
75
  "output": [
102
- "text",
103
- "image"
76
+ "text"
104
77
  ]
105
78
  },
106
79
  "open_weights": false,
107
80
  "limit": {
108
- "context": 131072,
109
- "output": 32768
81
+ "context": 1048576,
82
+ "output": 65536
110
83
  },
111
84
  "cost": {
112
85
  "input": 2,
113
- "output": 120
114
- }
115
- },
116
- "gemini-3-pro-image-preview": {
117
- "id": "gemini-3-pro-image-preview",
118
- "name": "Nano Banana Pro",
119
- "description": "Nano Banana Pro for higher-fidelity image generation and design-heavy edits",
120
- "family": "gemini-pro",
121
- "attachment": true,
122
- "reasoning": true,
123
- "reasoning_options": [],
124
- "tool_call": false,
125
- "temperature": true,
126
- "knowledge": "2025-01",
127
- "release_date": "2025-11-20",
128
- "last_updated": "2025-11-20",
129
- "modalities": {
130
- "input": [
131
- "text",
132
- "image"
86
+ "output": 12,
87
+ "cache_read": 0.2,
88
+ "tiers": [
89
+ {
90
+ "input": 4,
91
+ "output": 18,
92
+ "cache_read": 0.4,
93
+ "tier": {
94
+ "type": "context",
95
+ "size": 200000
96
+ }
97
+ }
133
98
  ],
134
- "output": [
135
- "text",
136
- "image"
137
- ]
138
- },
139
- "open_weights": false,
140
- "limit": {
141
- "context": 131072,
142
- "output": 32768
143
- },
144
- "cost": {
145
- "input": 2,
146
- "output": 120
99
+ "context_over_200k": {
100
+ "input": 4,
101
+ "output": 18,
102
+ "cache_read": 0.4
103
+ }
147
104
  }
148
105
  },
149
- "gemini-flash-lite-latest": {
150
- "id": "gemini-flash-lite-latest",
151
- "name": "Gemini Flash-Lite Latest",
152
- "description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
106
+ "gemini-3.1-flash-lite-image": {
107
+ "id": "gemini-3.1-flash-lite-image",
108
+ "name": "Nano Banana 2 Lite",
109
+ "description": "Fastest, most cost-efficient Gemini image model for high-volume 1K generation and editing",
153
110
  "family": "gemini-flash-lite",
154
111
  "attachment": true,
155
112
  "reasoning": true,
@@ -158,76 +115,56 @@
158
115
  "type": "effort",
159
116
  "values": [
160
117
  "minimal",
161
- "low",
162
- "medium",
163
118
  "high"
164
119
  ]
165
120
  }
166
121
  ],
167
122
  "tool_call": true,
168
- "structured_output": true,
123
+ "structured_output": false,
169
124
  "temperature": true,
170
125
  "knowledge": "2025-01",
171
- "release_date": "2026-05-07",
172
- "last_updated": "2026-05-07",
126
+ "release_date": "2026-06-30",
127
+ "last_updated": "2026-06-30",
173
128
  "modalities": {
174
129
  "input": [
175
130
  "text",
176
- "image",
177
- "video",
178
- "audio",
179
- "pdf"
131
+ "image"
180
132
  ],
181
133
  "output": [
182
- "text"
134
+ "text",
135
+ "image"
183
136
  ]
184
137
  },
185
138
  "open_weights": false,
186
139
  "limit": {
187
- "context": 1048576,
140
+ "context": 65536,
188
141
  "output": 65536
189
142
  },
190
143
  "cost": {
191
144
  "input": 0.25,
192
- "output": 1.5,
193
- "cache_read": 0.025,
194
- "input_audio": 0.5
145
+ "output": 30
195
146
  }
196
147
  },
197
- "gemini-3.5-flash-lite": {
198
- "id": "gemini-3.5-flash-lite",
199
- "name": "Gemini 3.5 Flash Lite",
200
- "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
201
- "family": "gemini-flash-lite",
148
+ "lyria-3-clip-preview": {
149
+ "id": "lyria-3-clip-preview",
150
+ "name": "Lyria 3 Clip Preview",
151
+ "description": "Music generation model for short 30-second clips, loops, and previews from text or image prompts",
152
+ "family": "lyria",
202
153
  "attachment": true,
203
- "reasoning": true,
204
- "reasoning_options": [
205
- {
206
- "type": "effort",
207
- "values": [
208
- "minimal",
209
- "low",
210
- "medium",
211
- "high"
212
- ]
213
- }
214
- ],
215
- "tool_call": true,
216
- "structured_output": true,
154
+ "reasoning": false,
155
+ "tool_call": false,
156
+ "structured_output": false,
217
157
  "temperature": true,
218
- "knowledge": "2026-03",
219
- "release_date": "2026-07-21",
220
- "last_updated": "2026-07-21",
158
+ "release_date": "2026-03-25",
159
+ "last_updated": "2026-03-25",
221
160
  "modalities": {
222
161
  "input": [
223
162
  "text",
224
- "image",
225
- "video",
226
- "audio",
227
- "pdf"
163
+ "image"
228
164
  ],
229
165
  "output": [
230
- "text"
166
+ "text",
167
+ "audio"
231
168
  ]
232
169
  },
233
170
  "open_weights": false,
@@ -236,33 +173,23 @@
236
173
  "output": 65536
237
174
  },
238
175
  "cost": {
239
- "input": 0.3,
240
- "output": 2.5,
241
- "cache_read": 0.03
176
+ "input": 0,
177
+ "output": 0
242
178
  }
243
179
  },
244
- "gemini-3.1-flash-lite-image": {
245
- "id": "gemini-3.1-flash-lite-image",
246
- "name": "Nano Banana 2 Lite",
247
- "description": "Fastest, most cost-efficient Gemini image model for high-volume 1K generation and editing",
248
- "family": "gemini-flash-lite",
180
+ "gemini-2.5-flash-image": {
181
+ "id": "gemini-2.5-flash-image",
182
+ "name": "Nano Banana",
183
+ "description": "Nano Banana image model for fast generation, edits, and character-consistent assets",
184
+ "family": "gemini-flash",
249
185
  "attachment": true,
250
186
  "reasoning": true,
251
- "reasoning_options": [
252
- {
253
- "type": "effort",
254
- "values": [
255
- "minimal",
256
- "high"
257
- ]
258
- }
259
- ],
260
- "tool_call": true,
261
- "structured_output": false,
187
+ "reasoning_options": [],
188
+ "tool_call": false,
262
189
  "temperature": true,
263
- "knowledge": "2025-01",
264
- "release_date": "2026-06-30",
265
- "last_updated": "2026-06-30",
190
+ "knowledge": "2024-06",
191
+ "release_date": "2025-08-26",
192
+ "last_updated": "2025-08-26",
266
193
  "modalities": {
267
194
  "input": [
268
195
  "text",
@@ -275,37 +202,28 @@
275
202
  },
276
203
  "open_weights": false,
277
204
  "limit": {
278
- "context": 65536,
279
- "output": 65536
205
+ "context": 32768,
206
+ "output": 32768
280
207
  },
281
208
  "cost": {
282
- "input": 0.25,
283
- "output": 30
209
+ "input": 0.3,
210
+ "output": 30,
211
+ "cache_read": 0.075
284
212
  }
285
213
  },
286
- "gemini-3.1-pro-preview": {
287
- "id": "gemini-3.1-pro-preview",
288
- "name": "Gemini 3.1 Pro Preview",
289
- "description": "Reasoning-first Gemini preview for agentic coding and complex problem solving",
214
+ "deep-research-max-preview-04-2026": {
215
+ "id": "deep-research-max-preview-04-2026",
216
+ "name": "Deep Research Max Preview (Apr-21-2026)",
217
+ "description": "Maximum-comprehensiveness agentic researcher for multi-step investigation, synthesis, and cited reports",
290
218
  "family": "gemini-pro",
291
219
  "attachment": true,
292
220
  "reasoning": true,
293
- "reasoning_options": [
294
- {
295
- "type": "effort",
296
- "values": [
297
- "low",
298
- "medium",
299
- "high"
300
- ]
301
- }
302
- ],
221
+ "reasoning_options": [],
303
222
  "tool_call": true,
304
- "structured_output": true,
305
223
  "temperature": true,
306
224
  "knowledge": "2025-01",
307
- "release_date": "2026-02-19",
308
- "last_updated": "2026-02-19",
225
+ "release_date": "2026-04-21",
226
+ "last_updated": "2026-04-21",
309
227
  "modalities": {
310
228
  "input": [
311
229
  "text",
@@ -315,12 +233,13 @@
315
233
  "pdf"
316
234
  ],
317
235
  "output": [
318
- "text"
236
+ "text",
237
+ "image"
319
238
  ]
320
239
  },
321
240
  "open_weights": false,
322
241
  "limit": {
323
- "context": 1048576,
242
+ "context": 131072,
324
243
  "output": 65536
325
244
  },
326
245
  "cost": {
@@ -345,10 +264,51 @@
345
264
  }
346
265
  }
347
266
  },
348
- "gemini-3.1-pro-preview-customtools": {
349
- "id": "gemini-3.1-pro-preview-customtools",
350
- "name": "Gemini 3.1 Pro Preview Custom Tools",
351
- "description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
267
+ "gemini-3-pro-image": {
268
+ "id": "gemini-3-pro-image",
269
+ "name": "Nano Banana Pro",
270
+ "description": "Nano Banana Pro for higher-fidelity image generation and design-heavy edits",
271
+ "family": "gemini-pro",
272
+ "attachment": true,
273
+ "reasoning": true,
274
+ "reasoning_options": [
275
+ {
276
+ "type": "effort",
277
+ "values": [
278
+ "low",
279
+ "high"
280
+ ]
281
+ }
282
+ ],
283
+ "tool_call": false,
284
+ "temperature": true,
285
+ "knowledge": "2025-01",
286
+ "release_date": "2026-05-28",
287
+ "last_updated": "2026-05-28",
288
+ "modalities": {
289
+ "input": [
290
+ "text",
291
+ "image"
292
+ ],
293
+ "output": [
294
+ "text",
295
+ "image"
296
+ ]
297
+ },
298
+ "open_weights": false,
299
+ "limit": {
300
+ "context": 131072,
301
+ "output": 32768
302
+ },
303
+ "cost": {
304
+ "input": 2,
305
+ "output": 120
306
+ }
307
+ },
308
+ "gemini-3.1-pro-preview": {
309
+ "id": "gemini-3.1-pro-preview",
310
+ "name": "Gemini 3.1 Pro Preview",
311
+ "description": "Reasoning-first Gemini preview for agentic coding and complex problem solving",
352
312
  "family": "gemini-pro",
353
313
  "attachment": true,
354
314
  "reasoning": true,
@@ -507,34 +467,131 @@
507
467
  "input_audio": 0.3
508
468
  }
509
469
  },
510
- "gemini-robotics-er-1.6-preview": {
511
- "id": "gemini-robotics-er-1.6-preview",
512
- "name": "Gemini Robotics-ER 1.6 Preview",
513
- "description": "Vision-language model for embodied reasoning: spatial understanding, task planning, and physical-world agentic robotics",
514
- "family": "gemini",
470
+ "gemini-2.5-computer-use-preview-10-2025": {
471
+ "id": "gemini-2.5-computer-use-preview-10-2025",
472
+ "name": "Gemini 2.5 Computer Use Preview 10-2025",
473
+ "description": "Specialized Gemini 2.5 model for browser-control agents that automate UI tasks",
474
+ "family": "gemini-pro",
475
+ "attachment": true,
476
+ "reasoning": true,
477
+ "reasoning_options": [],
478
+ "tool_call": true,
479
+ "temperature": true,
480
+ "knowledge": "2025-01",
481
+ "release_date": "2025-10-07",
482
+ "last_updated": "2025-10-07",
483
+ "modalities": {
484
+ "input": [
485
+ "text",
486
+ "image"
487
+ ],
488
+ "output": [
489
+ "text"
490
+ ]
491
+ },
492
+ "open_weights": false,
493
+ "limit": {
494
+ "context": 131072,
495
+ "output": 65536
496
+ },
497
+ "cost": {
498
+ "input": 1.25,
499
+ "output": 10,
500
+ "tiers": [
501
+ {
502
+ "input": 2.5,
503
+ "output": 15,
504
+ "tier": {
505
+ "type": "context",
506
+ "size": 200000
507
+ }
508
+ }
509
+ ],
510
+ "context_over_200k": {
511
+ "input": 2.5,
512
+ "output": 15
513
+ }
514
+ }
515
+ },
516
+ "gemini-3.6-flash": {
517
+ "id": "gemini-3.6-flash",
518
+ "name": "Gemini 3.6 Flash",
519
+ "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
520
+ "family": "gemini-flash",
515
521
  "attachment": true,
516
522
  "reasoning": true,
517
523
  "reasoning_options": [
518
524
  {
519
- "type": "toggle"
520
- },
525
+ "type": "effort",
526
+ "values": [
527
+ "minimal",
528
+ "low",
529
+ "medium",
530
+ "high"
531
+ ]
532
+ }
533
+ ],
534
+ "tool_call": true,
535
+ "structured_output": true,
536
+ "temperature": true,
537
+ "knowledge": "2026-03",
538
+ "release_date": "2026-07-21",
539
+ "last_updated": "2026-07-21",
540
+ "modalities": {
541
+ "input": [
542
+ "text",
543
+ "image",
544
+ "video",
545
+ "audio",
546
+ "pdf"
547
+ ],
548
+ "output": [
549
+ "text"
550
+ ]
551
+ },
552
+ "open_weights": false,
553
+ "limit": {
554
+ "context": 1048576,
555
+ "output": 65536
556
+ },
557
+ "cost": {
558
+ "input": 0.75,
559
+ "output": 3.75,
560
+ "cache_read": 0.075,
561
+ "input_audio": 0.75
562
+ }
563
+ },
564
+ "gemini-3.1-flash-lite": {
565
+ "id": "gemini-3.1-flash-lite",
566
+ "name": "Gemini 3.1 Flash Lite",
567
+ "description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
568
+ "family": "gemini-flash-lite",
569
+ "attachment": true,
570
+ "reasoning": true,
571
+ "reasoning_options": [
521
572
  {
522
- "type": "budget_tokens",
523
- "min": 0
573
+ "type": "effort",
574
+ "values": [
575
+ "minimal",
576
+ "low",
577
+ "medium",
578
+ "high"
579
+ ]
524
580
  }
525
581
  ],
526
582
  "tool_call": true,
527
583
  "structured_output": true,
528
584
  "temperature": true,
529
585
  "knowledge": "2025-01",
530
- "release_date": "2026-04-14",
531
- "last_updated": "2026-04-14",
586
+ "release_date": "2026-05-07",
587
+ "last_updated": "2026-05-07",
532
588
  "modalities": {
533
589
  "input": [
534
590
  "text",
535
591
  "image",
536
592
  "video",
537
- "audio"
593
+ "audio",
594
+ "pdf"
538
595
  ],
539
596
  "output": [
540
597
  "text"
@@ -542,64 +599,125 @@
542
599
  },
543
600
  "open_weights": false,
544
601
  "limit": {
545
- "context": 131072,
602
+ "context": 1048576,
546
603
  "output": 65536
547
604
  },
548
605
  "cost": {
549
- "input": 1,
550
- "output": 5,
551
- "input_audio": 2
606
+ "input": 0.25,
607
+ "output": 1.5,
608
+ "cache_read": 0.025,
609
+ "input_audio": 0.5
552
610
  }
553
611
  },
554
- "gemma-4-31b-it": {
555
- "id": "gemma-4-31b-it",
556
- "name": "Gemma 4 31B IT",
557
- "description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
558
- "family": "gemma",
612
+ "gemini-3.1-flash-live-preview": {
613
+ "id": "gemini-3.1-flash-live-preview",
614
+ "name": "Gemini 3.1 Flash Live Preview",
615
+ "description": "High-quality, low-latency Live API model for real-time dialogue and voice-first AI applications",
616
+ "family": "gemini-flash",
559
617
  "attachment": true,
560
618
  "reasoning": true,
561
619
  "reasoning_options": [
562
620
  {
563
- "type": "toggle"
621
+ "type": "effort",
622
+ "values": [
623
+ "minimal",
624
+ "low",
625
+ "medium",
626
+ "high"
627
+ ]
564
628
  }
565
629
  ],
566
630
  "tool_call": true,
567
- "structured_output": true,
631
+ "structured_output": false,
568
632
  "temperature": true,
569
- "release_date": "2026-04-02",
570
- "last_updated": "2026-04-02",
633
+ "knowledge": "2025-01",
634
+ "release_date": "2026-03-26",
635
+ "last_updated": "2026-03-26",
571
636
  "modalities": {
572
637
  "input": [
573
638
  "text",
574
- "image"
639
+ "image",
640
+ "video",
641
+ "audio"
575
642
  ],
576
643
  "output": [
644
+ "text",
645
+ "audio"
646
+ ]
647
+ },
648
+ "open_weights": false,
649
+ "limit": {
650
+ "context": 131072,
651
+ "output": 65536
652
+ },
653
+ "cost": {
654
+ "input": 0.75,
655
+ "output": 4.5,
656
+ "input_audio": 3,
657
+ "output_audio": 12
658
+ }
659
+ },
660
+ "gemini-2.5-pro-preview-tts": {
661
+ "id": "gemini-2.5-pro-preview-tts",
662
+ "name": "Gemini 2.5 Pro Preview TTS",
663
+ "description": "Speech generation model for controllable voice, narration, and audio delivery",
664
+ "family": "gemini-flash",
665
+ "attachment": false,
666
+ "reasoning": false,
667
+ "tool_call": false,
668
+ "temperature": true,
669
+ "knowledge": "2025-01",
670
+ "release_date": "2025-05-01",
671
+ "last_updated": "2025-05-01",
672
+ "modalities": {
673
+ "input": [
577
674
  "text"
675
+ ],
676
+ "output": [
677
+ "audio"
578
678
  ]
579
679
  },
580
- "open_weights": true,
680
+ "open_weights": false,
581
681
  "limit": {
582
- "context": 262144,
583
- "output": 32768
682
+ "context": 8192,
683
+ "output": 16384
684
+ },
685
+ "cost": {
686
+ "input": 1,
687
+ "output": 20
584
688
  }
585
689
  },
586
- "gemini-2.5-computer-use-preview-10-2025": {
587
- "id": "gemini-2.5-computer-use-preview-10-2025",
588
- "name": "Gemini 2.5 Computer Use Preview 10-2025",
589
- "description": "Specialized Gemini 2.5 model for browser-control agents that automate UI tasks",
590
- "family": "gemini-pro",
690
+ "gemini-3.5-flash": {
691
+ "id": "gemini-3.5-flash",
692
+ "name": "Gemini 3.5 Flash",
693
+ "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
694
+ "family": "gemini-flash",
591
695
  "attachment": true,
592
696
  "reasoning": true,
593
- "reasoning_options": [],
697
+ "reasoning_options": [
698
+ {
699
+ "type": "effort",
700
+ "values": [
701
+ "minimal",
702
+ "low",
703
+ "medium",
704
+ "high"
705
+ ]
706
+ }
707
+ ],
594
708
  "tool_call": true,
709
+ "structured_output": true,
595
710
  "temperature": true,
596
711
  "knowledge": "2025-01",
597
- "release_date": "2025-10-07",
598
- "last_updated": "2025-10-07",
712
+ "release_date": "2026-05-19",
713
+ "last_updated": "2026-05-19",
599
714
  "modalities": {
600
715
  "input": [
601
716
  "text",
602
- "image"
717
+ "image",
718
+ "video",
719
+ "audio",
720
+ "pdf"
603
721
  ],
604
722
  "output": [
605
723
  "text"
@@ -607,48 +725,77 @@
607
725
  },
608
726
  "open_weights": false,
609
727
  "limit": {
610
- "context": 131072,
611
- "output": 65536
728
+ "context": 1048576,
729
+ "output": 65536
730
+ },
731
+ "cost": {
732
+ "input": 1.5,
733
+ "output": 9,
734
+ "cache_read": 0.15,
735
+ "input_audio": 1.5
736
+ }
737
+ },
738
+ "veo-3.1-generate-preview": {
739
+ "id": "veo-3.1-generate-preview",
740
+ "name": "Veo 3.1",
741
+ "description": "Video model for prompt-guided generation, editing, and motion workflows",
742
+ "family": "veo",
743
+ "attachment": true,
744
+ "reasoning": false,
745
+ "tool_call": false,
746
+ "temperature": false,
747
+ "release_date": "2025-10-15",
748
+ "last_updated": "2026-01",
749
+ "modalities": {
750
+ "input": [
751
+ "text",
752
+ "image"
753
+ ],
754
+ "output": [
755
+ "video"
756
+ ]
757
+ },
758
+ "open_weights": false,
759
+ "limit": {
760
+ "context": 480,
761
+ "output": 8192
612
762
  },
613
- "cost": {
614
- "input": 1.25,
615
- "output": 10,
616
- "tiers": [
617
- {
618
- "input": 2.5,
619
- "output": 15,
620
- "tier": {
621
- "type": "context",
622
- "size": 200000
623
- }
624
- }
625
- ],
626
- "context_over_200k": {
627
- "input": 2.5,
628
- "output": 15
629
- }
630
- }
763
+ "status": "beta"
631
764
  },
632
- "lyria-3-clip-preview": {
633
- "id": "lyria-3-clip-preview",
634
- "name": "Lyria 3 Clip Preview",
635
- "description": "Music generation model for short 30-second clips, loops, and previews from text or image prompts",
636
- "family": "lyria",
765
+ "gemini-3.1-flash-lite-preview": {
766
+ "id": "gemini-3.1-flash-lite-preview",
767
+ "name": "Gemini 3.1 Flash Lite Preview",
768
+ "description": "Legacy model retained for compatibility with older integrations",
769
+ "family": "gemini-flash-lite",
637
770
  "attachment": true,
638
- "reasoning": false,
639
- "tool_call": false,
640
- "structured_output": false,
771
+ "reasoning": true,
772
+ "reasoning_options": [
773
+ {
774
+ "type": "effort",
775
+ "values": [
776
+ "minimal",
777
+ "low",
778
+ "medium",
779
+ "high"
780
+ ]
781
+ }
782
+ ],
783
+ "tool_call": true,
784
+ "structured_output": true,
641
785
  "temperature": true,
642
- "release_date": "2026-03-25",
643
- "last_updated": "2026-03-25",
786
+ "knowledge": "2025-01",
787
+ "release_date": "2026-03-03",
788
+ "last_updated": "2026-03-03",
644
789
  "modalities": {
645
790
  "input": [
646
791
  "text",
647
- "image"
792
+ "image",
793
+ "video",
794
+ "audio",
795
+ "pdf"
648
796
  ],
649
797
  "output": [
650
- "text",
651
- "audio"
798
+ "text"
652
799
  ]
653
800
  },
654
801
  "open_weights": false,
@@ -656,9 +803,12 @@
656
803
  "context": 1048576,
657
804
  "output": 65536
658
805
  },
806
+ "status": "deprecated",
659
807
  "cost": {
660
- "input": 0,
661
- "output": 0
808
+ "input": 0.25,
809
+ "output": 1.5,
810
+ "cache_read": 0.025,
811
+ "input_audio": 0.5
662
812
  }
663
813
  },
664
814
  "veo-3.1-fast-generate-preview": {
@@ -688,70 +838,70 @@
688
838
  "output": 8192
689
839
  }
690
840
  },
691
- "gemini-embedding-001": {
692
- "id": "gemini-embedding-001",
693
- "name": "Gemini Embedding 001",
694
- "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
695
- "family": "gemini",
841
+ "gemini-2.5-flash-preview-tts": {
842
+ "id": "gemini-2.5-flash-preview-tts",
843
+ "name": "Gemini 2.5 Flash Preview TTS",
844
+ "description": "Speech generation model for controllable voice, narration, and audio delivery",
845
+ "family": "gemini-flash",
696
846
  "attachment": false,
697
847
  "reasoning": false,
698
848
  "tool_call": false,
699
- "temperature": false,
700
- "knowledge": "2025-05",
701
- "release_date": "2025-05-20",
702
- "last_updated": "2025-05-20",
849
+ "temperature": true,
850
+ "knowledge": "2025-01",
851
+ "release_date": "2025-05-01",
852
+ "last_updated": "2025-05-01",
703
853
  "modalities": {
704
854
  "input": [
705
855
  "text"
706
856
  ],
707
857
  "output": [
708
- "text"
858
+ "audio"
709
859
  ]
710
860
  },
711
861
  "open_weights": false,
712
862
  "limit": {
713
- "context": 2048,
714
- "output": 1
863
+ "context": 8192,
864
+ "output": 16384
715
865
  },
716
866
  "cost": {
717
- "input": 0.15,
718
- "output": 0
867
+ "input": 0.5,
868
+ "output": 10
719
869
  }
720
870
  },
721
- "gemini-2.5-pro-preview-tts": {
722
- "id": "gemini-2.5-pro-preview-tts",
723
- "name": "Gemini 2.5 Pro Preview TTS",
724
- "description": "Speech generation model for controllable voice, narration, and audio delivery",
725
- "family": "gemini-flash",
871
+ "gemini-embedding-001": {
872
+ "id": "gemini-embedding-001",
873
+ "name": "Gemini Embedding 001",
874
+ "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
875
+ "family": "gemini",
726
876
  "attachment": false,
727
877
  "reasoning": false,
728
878
  "tool_call": false,
729
- "temperature": true,
730
- "knowledge": "2025-01",
731
- "release_date": "2025-05-01",
732
- "last_updated": "2025-05-01",
879
+ "temperature": false,
880
+ "knowledge": "2025-05",
881
+ "release_date": "2025-05-20",
882
+ "last_updated": "2025-05-20",
733
883
  "modalities": {
734
884
  "input": [
735
885
  "text"
736
886
  ],
737
887
  "output": [
738
- "audio"
888
+ "text"
739
889
  ]
740
890
  },
741
891
  "open_weights": false,
742
892
  "limit": {
743
- "context": 8192,
744
- "output": 16384
893
+ "context": 2048,
894
+ "output": 1
745
895
  },
746
896
  "cost": {
747
- "input": 1,
748
- "output": 20
897
+ "input": 0.15,
898
+ "output": 0
749
899
  }
750
900
  },
751
- "gemini-flash-latest": {
752
- "id": "gemini-flash-latest",
753
- "name": "Gemini Flash Latest",
754
- "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
901
+ "gemini-3.1-flash-image": {
902
+ "id": "gemini-3.1-flash-image",
903
+ "name": "Nano Banana 2",
904
+ "description": "Image model for prompt-driven generation, editing, and visual design workflows",
755
905
  "family": "gemini-flash",
756
906
  "attachment": true,
757
907
  "reasoning": true,
@@ -760,47 +910,42 @@
760
910
  "type": "effort",
761
911
  "values": [
762
912
  "minimal",
763
- "low",
764
- "medium",
765
913
  "high"
766
914
  ]
767
915
  }
768
916
  ],
769
- "tool_call": true,
770
- "structured_output": true,
917
+ "tool_call": false,
771
918
  "temperature": true,
772
919
  "knowledge": "2025-01",
773
- "release_date": "2026-05-19",
774
- "last_updated": "2026-05-19",
920
+ "release_date": "2026-05-28",
921
+ "last_updated": "2026-05-28",
775
922
  "modalities": {
776
923
  "input": [
777
924
  "text",
778
925
  "image",
779
926
  "video",
780
- "audio",
781
927
  "pdf"
782
928
  ],
783
929
  "output": [
784
- "text"
930
+ "text",
931
+ "image"
785
932
  ]
786
933
  },
787
934
  "open_weights": false,
788
935
  "limit": {
789
- "context": 1048576,
936
+ "context": 65536,
790
937
  "output": 65536
791
938
  },
792
939
  "cost": {
793
- "input": 1.5,
794
- "output": 9,
795
- "cache_read": 0.15,
796
- "input_audio": 1.5
940
+ "input": 0.5,
941
+ "output": 60
797
942
  }
798
943
  },
799
- "gemini-3.5-flash": {
800
- "id": "gemini-3.5-flash",
801
- "name": "Gemini 3.5 Flash",
944
+ "gemini-3.5-flash-lite": {
945
+ "id": "gemini-3.5-flash-lite",
946
+ "name": "Gemini 3.5 Flash Lite",
802
947
  "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
803
- "family": "gemini-flash",
948
+ "family": "gemini-flash-lite",
804
949
  "attachment": true,
805
950
  "reasoning": true,
806
951
  "reasoning_options": [
@@ -817,9 +962,9 @@
817
962
  "tool_call": true,
818
963
  "structured_output": true,
819
964
  "temperature": true,
820
- "knowledge": "2025-01",
821
- "release_date": "2026-05-19",
822
- "last_updated": "2026-05-19",
965
+ "knowledge": "2026-03",
966
+ "release_date": "2026-07-21",
967
+ "last_updated": "2026-07-21",
823
968
  "modalities": {
824
969
  "input": [
825
970
  "text",
@@ -838,88 +983,90 @@
838
983
  "output": 65536
839
984
  },
840
985
  "cost": {
841
- "input": 1.5,
842
- "output": 9,
843
- "cache_read": 0.15,
844
- "input_audio": 1.5
986
+ "input": 0.3,
987
+ "output": 2.5,
988
+ "cache_read": 0.03
845
989
  }
846
990
  },
847
- "gemini-3.5-live-translate-preview": {
848
- "id": "gemini-3.5-live-translate-preview",
849
- "name": "Gemini 3.5 Live Translate Preview",
850
- "description": "Low-latency audio-to-audio model for real-time speech translation across 70+ languages",
991
+ "gemini-3-pro-image-preview": {
992
+ "id": "gemini-3-pro-image-preview",
993
+ "name": "Nano Banana Pro",
994
+ "description": "Nano Banana Pro for higher-fidelity image generation and design-heavy edits",
851
995
  "family": "gemini-pro",
852
- "attachment": false,
853
- "reasoning": false,
996
+ "attachment": true,
997
+ "reasoning": true,
998
+ "reasoning_options": [],
854
999
  "tool_call": false,
855
1000
  "temperature": true,
856
1001
  "knowledge": "2025-01",
857
- "release_date": "2026-06-09",
858
- "last_updated": "2026-06-09",
1002
+ "release_date": "2025-11-20",
1003
+ "last_updated": "2025-11-20",
859
1004
  "modalities": {
860
1005
  "input": [
861
- "audio"
1006
+ "text",
1007
+ "image"
862
1008
  ],
863
1009
  "output": [
864
- "audio",
865
- "text"
1010
+ "text",
1011
+ "image"
866
1012
  ]
867
1013
  },
868
1014
  "open_weights": false,
869
1015
  "limit": {
870
- "context": 16384,
1016
+ "context": 131072,
871
1017
  "output": 32768
872
1018
  },
873
1019
  "cost": {
874
- "input": 3.5,
875
- "output": 21,
876
- "input_audio": 3.5,
877
- "output_audio": 21
1020
+ "input": 2,
1021
+ "output": 120
878
1022
  }
879
1023
  },
880
- "veo-3.1-lite-generate-preview": {
881
- "id": "veo-3.1-lite-generate-preview",
882
- "name": "Veo 3.1 lite",
883
- "description": "Video model for prompt-guided generation, editing, and motion workflows",
884
- "family": "veo",
885
- "attachment": true,
886
- "reasoning": false,
1024
+ "gemini-3.1-flash-tts-preview": {
1025
+ "id": "gemini-3.1-flash-tts-preview",
1026
+ "name": "Gemini 3.1 Flash TTS Preview",
1027
+ "description": "Low-latency speech generation with steerable prompts and expressive audio tags",
1028
+ "family": "gemini-flash",
1029
+ "attachment": false,
1030
+ "reasoning": true,
1031
+ "reasoning_options": [],
887
1032
  "tool_call": false,
888
- "temperature": false,
889
- "release_date": "2026-03-31",
890
- "last_updated": "2026-03-31",
1033
+ "temperature": true,
1034
+ "knowledge": "2025-01",
1035
+ "release_date": "2026-04-15",
1036
+ "last_updated": "2026-04-15",
891
1037
  "modalities": {
892
1038
  "input": [
893
- "text",
894
- "image"
1039
+ "text"
895
1040
  ],
896
1041
  "output": [
897
- "video"
1042
+ "audio"
898
1043
  ]
899
1044
  },
900
1045
  "open_weights": false,
901
1046
  "limit": {
902
- "context": 480,
903
- "output": 8192
1047
+ "context": 8192,
1048
+ "output": 16384
1049
+ },
1050
+ "cost": {
1051
+ "input": 1,
1052
+ "output": 20
904
1053
  }
905
1054
  },
906
- "gemini-omni-flash-preview": {
907
- "id": "gemini-omni-flash-preview",
908
- "name": "Gemini Omni Flash Preview",
909
- "description": "Video generation and editing model for fast, conversational text- and image-to-video workflows",
910
- "family": "gemini",
1055
+ "veo-3.1-lite-generate-preview": {
1056
+ "id": "veo-3.1-lite-generate-preview",
1057
+ "name": "Veo 3.1 lite",
1058
+ "description": "Video model for prompt-guided generation, editing, and motion workflows",
1059
+ "family": "veo",
911
1060
  "attachment": true,
912
- "reasoning": true,
913
- "reasoning_options": [],
1061
+ "reasoning": false,
914
1062
  "tool_call": false,
915
- "temperature": true,
916
- "release_date": "2026-06-30",
917
- "last_updated": "2026-06-30",
1063
+ "temperature": false,
1064
+ "release_date": "2026-03-31",
1065
+ "last_updated": "2026-03-31",
918
1066
  "modalities": {
919
1067
  "input": [
920
1068
  "text",
921
- "image",
922
- "video"
1069
+ "image"
923
1070
  ],
924
1071
  "output": [
925
1072
  "video"
@@ -927,18 +1074,14 @@
927
1074
  },
928
1075
  "open_weights": false,
929
1076
  "limit": {
930
- "context": 131072,
931
- "output": 65536
932
- },
933
- "cost": {
934
- "input": 1.5,
935
- "output": 17.5
1077
+ "context": 480,
1078
+ "output": 8192
936
1079
  }
937
1080
  },
938
- "gemini-3.1-flash-lite": {
939
- "id": "gemini-3.1-flash-lite",
940
- "name": "Gemini 3.1 Flash Lite",
941
- "description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
1081
+ "gemini-flash-lite-latest": {
1082
+ "id": "gemini-flash-lite-latest",
1083
+ "name": "Gemini Flash-Lite Latest",
1084
+ "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
942
1085
  "family": "gemini-flash-lite",
943
1086
  "attachment": true,
944
1087
  "reasoning": true,
@@ -956,9 +1099,9 @@
956
1099
  "tool_call": true,
957
1100
  "structured_output": true,
958
1101
  "temperature": true,
959
- "knowledge": "2025-01",
960
- "release_date": "2026-05-07",
961
- "last_updated": "2026-05-07",
1102
+ "knowledge": "2026-03",
1103
+ "release_date": "2026-07-21",
1104
+ "last_updated": "2026-07-21",
962
1105
  "modalities": {
963
1106
  "input": [
964
1107
  "text",
@@ -977,116 +1120,43 @@
977
1120
  "output": 65536
978
1121
  },
979
1122
  "cost": {
980
- "input": 0.25,
981
- "output": 1.5,
982
- "cache_read": 0.025,
983
- "input_audio": 0.5
1123
+ "input": 0.3,
1124
+ "output": 2.5,
1125
+ "cache_read": 0.03
984
1126
  }
985
1127
  },
986
- "gemini-2.5-flash": {
987
- "id": "gemini-2.5-flash",
988
- "name": "Gemini 2.5 Flash",
989
- "description": "Fast Gemini workhorse for multimodal apps where latency and price matter",
990
- "family": "gemini-flash",
1128
+ "gemma-4-31b-it": {
1129
+ "id": "gemma-4-31b-it",
1130
+ "name": "Gemma 4 31B IT",
1131
+ "description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
1132
+ "family": "gemma",
991
1133
  "attachment": true,
992
1134
  "reasoning": true,
993
1135
  "reasoning_options": [
994
1136
  {
995
1137
  "type": "toggle"
996
- },
997
- {
998
- "type": "budget_tokens",
999
- "min": 0,
1000
- "max": 24576
1001
1138
  }
1002
1139
  ],
1003
1140
  "tool_call": true,
1004
1141
  "structured_output": true,
1005
1142
  "temperature": true,
1006
- "knowledge": "2025-01",
1007
- "release_date": "2025-06-17",
1008
- "last_updated": "2025-06-17",
1143
+ "release_date": "2026-04-02",
1144
+ "last_updated": "2026-04-02",
1009
1145
  "modalities": {
1010
1146
  "input": [
1011
1147
  "text",
1012
- "image",
1013
- "audio",
1014
- "video",
1015
- "pdf"
1148
+ "image"
1016
1149
  ],
1017
1150
  "output": [
1018
1151
  "text"
1019
1152
  ]
1020
1153
  },
1021
- "open_weights": false,
1022
- "limit": {
1023
- "context": 1048576,
1024
- "output": 65536
1025
- },
1026
- "cost": {
1027
- "input": 0.3,
1028
- "output": 2.5,
1029
- "cache_read": 0.03,
1030
- "input_audio": 1
1031
- }
1032
- },
1033
- "gemini-2.5-flash-preview-tts": {
1034
- "id": "gemini-2.5-flash-preview-tts",
1035
- "name": "Gemini 2.5 Flash Preview TTS",
1036
- "description": "Speech generation model for controllable voice, narration, and audio delivery",
1037
- "family": "gemini-flash",
1038
- "attachment": false,
1039
- "reasoning": false,
1040
- "tool_call": false,
1041
- "temperature": true,
1042
- "knowledge": "2025-01",
1043
- "release_date": "2025-05-01",
1044
- "last_updated": "2025-05-01",
1045
- "modalities": {
1046
- "input": [
1047
- "text"
1048
- ],
1049
- "output": [
1050
- "audio"
1051
- ]
1052
- },
1053
- "open_weights": false,
1154
+ "open_weights": true,
1054
1155
  "limit": {
1055
- "context": 8192,
1056
- "output": 16384
1057
- },
1058
- "cost": {
1059
- "input": 0.5,
1060
- "output": 10
1156
+ "context": 262144,
1157
+ "output": 32768
1061
1158
  }
1062
1159
  },
1063
- "veo-3.1-generate-preview": {
1064
- "id": "veo-3.1-generate-preview",
1065
- "name": "Veo 3.1",
1066
- "description": "Video model for prompt-guided generation, editing, and motion workflows",
1067
- "family": "veo",
1068
- "attachment": true,
1069
- "reasoning": false,
1070
- "tool_call": false,
1071
- "temperature": false,
1072
- "release_date": "2025-10-15",
1073
- "last_updated": "2026-01",
1074
- "modalities": {
1075
- "input": [
1076
- "text",
1077
- "image"
1078
- ],
1079
- "output": [
1080
- "video"
1081
- ]
1082
- },
1083
- "open_weights": false,
1084
- "limit": {
1085
- "context": 480,
1086
- "output": 8192
1087
- },
1088
- "status": "beta"
1089
- },
1090
1160
  "gemini-embedding-2": {
1091
1161
  "id": "gemini-embedding-2",
1092
1162
  "name": "Gemini Embedding 2",
@@ -1122,10 +1192,10 @@
1122
1192
  "input_audio": 6.5
1123
1193
  }
1124
1194
  },
1125
- "gemini-3.6-flash": {
1126
- "id": "gemini-3.6-flash",
1127
- "name": "Gemini 3.6 Flash",
1128
- "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
1195
+ "gemini-3-flash-preview": {
1196
+ "id": "gemini-3-flash-preview",
1197
+ "name": "Gemini 3 Flash Preview",
1198
+ "description": "New Gemini flash lane bringing frontier-style multimodal reasoning to cheaper runs",
1129
1199
  "family": "gemini-flash",
1130
1200
  "attachment": true,
1131
1201
  "reasoning": true,
@@ -1143,9 +1213,9 @@
1143
1213
  "tool_call": true,
1144
1214
  "structured_output": true,
1145
1215
  "temperature": true,
1146
- "knowledge": "2026-03",
1147
- "release_date": "2026-07-21",
1148
- "last_updated": "2026-07-21",
1216
+ "knowledge": "2025-01",
1217
+ "release_date": "2025-12-17",
1218
+ "last_updated": "2025-12-17",
1149
1219
  "modalities": {
1150
1220
  "input": [
1151
1221
  "text",
@@ -1163,60 +1233,17 @@
1163
1233
  "context": 1048576,
1164
1234
  "output": 65536
1165
1235
  },
1166
- "cost": {
1167
- "input": 1.5,
1168
- "output": 7.5,
1169
- "cache_read": 0.15,
1170
- "input_audio": 1.5
1171
- }
1172
- },
1173
- "gemini-3.1-flash-image": {
1174
- "id": "gemini-3.1-flash-image",
1175
- "name": "Nano Banana 2",
1176
- "description": "Image model for prompt-driven generation, editing, and visual design workflows",
1177
- "family": "gemini-flash",
1178
- "attachment": true,
1179
- "reasoning": true,
1180
- "reasoning_options": [
1181
- {
1182
- "type": "effort",
1183
- "values": [
1184
- "minimal",
1185
- "high"
1186
- ]
1187
- }
1188
- ],
1189
- "tool_call": false,
1190
- "temperature": true,
1191
- "knowledge": "2025-01",
1192
- "release_date": "2026-05-28",
1193
- "last_updated": "2026-05-28",
1194
- "modalities": {
1195
- "input": [
1196
- "text",
1197
- "image",
1198
- "video",
1199
- "pdf"
1200
- ],
1201
- "output": [
1202
- "text",
1203
- "image"
1204
- ]
1205
- },
1206
- "open_weights": false,
1207
- "limit": {
1208
- "context": 65536,
1209
- "output": 65536
1210
- },
1211
1236
  "cost": {
1212
1237
  "input": 0.5,
1213
- "output": 60
1238
+ "output": 3,
1239
+ "cache_read": 0.05,
1240
+ "input_audio": 1
1214
1241
  }
1215
1242
  },
1216
- "gemini-3.1-flash-image-preview": {
1217
- "id": "gemini-3.1-flash-image-preview",
1218
- "name": "Nano Banana 2",
1219
- "description": "Image model for prompt-driven generation, editing, and visual design workflows",
1243
+ "gemini-3.8-flash": {
1244
+ "id": "gemini-3.8-flash",
1245
+ "name": "Gemini 3.8 Flash",
1246
+ "description": "Google's most intelligent Flash model, engineered for long-horizon software engineering, autonomous agents, and complex enterprise workflows",
1220
1247
  "family": "gemini-flash",
1221
1248
  "attachment": true,
1222
1249
  "reasoning": true,
@@ -1224,103 +1251,94 @@
1224
1251
  {
1225
1252
  "type": "effort",
1226
1253
  "values": [
1227
- "minimal",
1254
+ "low",
1255
+ "medium",
1228
1256
  "high"
1229
1257
  ]
1230
1258
  }
1231
1259
  ],
1232
- "tool_call": false,
1260
+ "tool_call": true,
1261
+ "structured_output": true,
1233
1262
  "temperature": true,
1234
- "knowledge": "2025-01",
1235
- "release_date": "2026-02-26",
1236
- "last_updated": "2026-02-26",
1263
+ "release_date": "2026-09-02",
1264
+ "last_updated": "2026-09-02",
1237
1265
  "modalities": {
1238
1266
  "input": [
1239
1267
  "text",
1240
1268
  "image",
1269
+ "video",
1270
+ "audio",
1241
1271
  "pdf"
1242
1272
  ],
1243
1273
  "output": [
1244
- "text",
1245
- "image"
1274
+ "text"
1246
1275
  ]
1247
1276
  },
1248
1277
  "open_weights": false,
1249
1278
  "limit": {
1250
- "context": 65536,
1279
+ "context": 1048576,
1251
1280
  "output": 65536
1252
1281
  },
1253
1282
  "cost": {
1254
- "input": 0.5,
1255
- "output": 60
1283
+ "input": 0.75,
1284
+ "output": 3.75,
1285
+ "cache_read": 0.075,
1286
+ "input_audio": 0.75
1256
1287
  }
1257
1288
  },
1258
- "lyria-3-pro-preview": {
1259
- "id": "lyria-3-pro-preview",
1260
- "name": "Lyria 3 Pro Preview",
1261
- "description": "Music generation model for full-length songs from text or images with vocals and structure",
1262
- "family": "lyria",
1263
- "attachment": true,
1289
+ "gemini-3.5-live-translate-preview": {
1290
+ "id": "gemini-3.5-live-translate-preview",
1291
+ "name": "Gemini 3.5 Live Translate Preview",
1292
+ "description": "Low-latency audio-to-audio model for real-time speech translation across 70+ languages",
1293
+ "family": "gemini-pro",
1294
+ "attachment": false,
1264
1295
  "reasoning": false,
1265
1296
  "tool_call": false,
1266
- "structured_output": false,
1267
1297
  "temperature": true,
1268
- "release_date": "2026-03-25",
1269
- "last_updated": "2026-03-25",
1298
+ "knowledge": "2025-01",
1299
+ "release_date": "2026-06-09",
1300
+ "last_updated": "2026-06-09",
1270
1301
  "modalities": {
1271
1302
  "input": [
1272
- "text",
1273
- "image"
1303
+ "audio"
1274
1304
  ],
1275
1305
  "output": [
1276
- "text",
1277
- "audio"
1306
+ "audio",
1307
+ "text"
1278
1308
  ]
1279
1309
  },
1280
1310
  "open_weights": false,
1281
1311
  "limit": {
1282
- "context": 1048576,
1283
- "output": 65536
1312
+ "context": 16384,
1313
+ "output": 32768
1284
1314
  },
1285
1315
  "cost": {
1286
- "input": 0,
1287
- "output": 0
1316
+ "input": 3.5,
1317
+ "output": 21,
1318
+ "input_audio": 3.5,
1319
+ "output_audio": 21
1288
1320
  }
1289
1321
  },
1290
- "gemini-3.1-flash-lite-preview": {
1291
- "id": "gemini-3.1-flash-lite-preview",
1292
- "name": "Gemini 3.1 Flash Lite Preview",
1293
- "description": "Legacy model retained for compatibility with older integrations",
1294
- "family": "gemini-flash-lite",
1322
+ "lyria-3-pro-preview": {
1323
+ "id": "lyria-3-pro-preview",
1324
+ "name": "Lyria 3 Pro Preview",
1325
+ "description": "Music generation model for full-length songs from text or images with vocals and structure",
1326
+ "family": "lyria",
1295
1327
  "attachment": true,
1296
- "reasoning": true,
1297
- "reasoning_options": [
1298
- {
1299
- "type": "effort",
1300
- "values": [
1301
- "minimal",
1302
- "low",
1303
- "medium",
1304
- "high"
1305
- ]
1306
- }
1307
- ],
1308
- "tool_call": true,
1309
- "structured_output": true,
1328
+ "reasoning": false,
1329
+ "tool_call": false,
1330
+ "structured_output": false,
1310
1331
  "temperature": true,
1311
- "knowledge": "2025-01",
1312
- "release_date": "2026-03-03",
1313
- "last_updated": "2026-03-03",
1332
+ "release_date": "2026-03-25",
1333
+ "last_updated": "2026-03-25",
1314
1334
  "modalities": {
1315
1335
  "input": [
1316
1336
  "text",
1317
- "image",
1318
- "video",
1319
- "audio",
1320
- "pdf"
1337
+ "image"
1321
1338
  ],
1322
1339
  "output": [
1323
- "text"
1340
+ "text",
1341
+ "audio"
1324
1342
  ]
1325
1343
  },
1326
1344
  "open_weights": false,
@@ -1328,18 +1346,15 @@
1328
1346
  "context": 1048576,
1329
1347
  "output": 65536
1330
1348
  },
1331
- "status": "deprecated",
1332
1349
  "cost": {
1333
- "input": 0.25,
1334
- "output": 1.5,
1335
- "cache_read": 0.025,
1336
- "input_audio": 0.5
1350
+ "input": 0,
1351
+ "output": 0
1337
1352
  }
1338
1353
  },
1339
- "gemini-3-flash-preview": {
1340
- "id": "gemini-3-flash-preview",
1341
- "name": "Gemini 3 Flash Preview",
1342
- "description": "New Gemini flash lane bringing frontier-style multimodal reasoning to cheaper runs",
1354
+ "gemini-3.7-flash": {
1355
+ "id": "gemini-3.7-flash",
1356
+ "name": "Gemini 3.7 Flash",
1357
+ "description": "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning",
1343
1358
  "family": "gemini-flash",
1344
1359
  "attachment": true,
1345
1360
  "reasoning": true,
@@ -1347,7 +1362,6 @@
1347
1362
  {
1348
1363
  "type": "effort",
1349
1364
  "values": [
1350
- "minimal",
1351
1365
  "low",
1352
1366
  "medium",
1353
1367
  "high"
@@ -1357,9 +1371,9 @@
1357
1371
  "tool_call": true,
1358
1372
  "structured_output": true,
1359
1373
  "temperature": true,
1360
- "knowledge": "2025-01",
1361
- "release_date": "2025-12-17",
1362
- "last_updated": "2025-12-17",
1374
+ "knowledge": "2026-03",
1375
+ "release_date": "2026-08-13",
1376
+ "last_updated": "2026-08-13",
1363
1377
  "modalities": {
1364
1378
  "input": [
1365
1379
  "text",
@@ -1378,44 +1392,10 @@
1378
1392
  "output": 65536
1379
1393
  },
1380
1394
  "cost": {
1381
- "input": 0.5,
1382
- "output": 3,
1383
- "cache_read": 0.05,
1384
- "input_audio": 1
1385
- }
1386
- },
1387
- "gemini-2.5-flash-image": {
1388
- "id": "gemini-2.5-flash-image",
1389
- "name": "Nano Banana",
1390
- "description": "Nano Banana image model for fast generation, edits, and character-consistent assets",
1391
- "family": "gemini-flash",
1392
- "attachment": true,
1393
- "reasoning": true,
1394
- "reasoning_options": [],
1395
- "tool_call": false,
1396
- "temperature": true,
1397
- "knowledge": "2024-06",
1398
- "release_date": "2025-08-26",
1399
- "last_updated": "2025-08-26",
1400
- "modalities": {
1401
- "input": [
1402
- "text",
1403
- "image"
1404
- ],
1405
- "output": [
1406
- "text",
1407
- "image"
1408
- ]
1409
- },
1410
- "open_weights": false,
1411
- "limit": {
1412
- "context": 32768,
1413
- "output": 32768
1414
- },
1415
- "cost": {
1416
- "input": 0.3,
1417
- "output": 30,
1418
- "cache_read": 0.075
1395
+ "input": 0.75,
1396
+ "output": 3.75,
1397
+ "cache_read": 0.075,
1398
+ "input_audio": 0.75
1419
1399
  }
1420
1400
  },
1421
1401
  "gemini-2.5-pro": {
@@ -1477,19 +1457,29 @@
1477
1457
  }
1478
1458
  }
1479
1459
  },
1480
- "deep-research-max-preview-04-2026": {
1481
- "id": "deep-research-max-preview-04-2026",
1482
- "name": "Deep Research Max Preview (Apr-21-2026)",
1483
- "description": "Maximum-comprehensiveness agentic researcher for multi-step investigation, synthesis, and cited reports",
1484
- "family": "gemini-pro",
1460
+ "gemini-flash-latest": {
1461
+ "id": "gemini-flash-latest",
1462
+ "name": "Gemini Flash Latest",
1463
+ "description": "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning",
1464
+ "family": "gemini-flash",
1485
1465
  "attachment": true,
1486
1466
  "reasoning": true,
1487
- "reasoning_options": [],
1467
+ "reasoning_options": [
1468
+ {
1469
+ "type": "effort",
1470
+ "values": [
1471
+ "low",
1472
+ "medium",
1473
+ "high"
1474
+ ]
1475
+ }
1476
+ ],
1488
1477
  "tool_call": true,
1478
+ "structured_output": true,
1489
1479
  "temperature": true,
1490
- "knowledge": "2025-01",
1491
- "release_date": "2026-04-21",
1492
- "last_updated": "2026-04-21",
1480
+ "knowledge": "2026-03",
1481
+ "release_date": "2026-08-13",
1482
+ "last_updated": "2026-08-13",
1493
1483
  "modalities": {
1494
1484
  "input": [
1495
1485
  "text",
@@ -1499,41 +1489,25 @@
1499
1489
  "pdf"
1500
1490
  ],
1501
1491
  "output": [
1502
- "text",
1503
- "image"
1492
+ "text"
1504
1493
  ]
1505
1494
  },
1506
1495
  "open_weights": false,
1507
1496
  "limit": {
1508
- "context": 131072,
1497
+ "context": 1048576,
1509
1498
  "output": 65536
1510
1499
  },
1511
1500
  "cost": {
1512
- "input": 2,
1513
- "output": 12,
1514
- "cache_read": 0.2,
1515
- "tiers": [
1516
- {
1517
- "input": 4,
1518
- "output": 18,
1519
- "cache_read": 0.4,
1520
- "tier": {
1521
- "type": "context",
1522
- "size": 200000
1523
- }
1524
- }
1525
- ],
1526
- "context_over_200k": {
1527
- "input": 4,
1528
- "output": 18,
1529
- "cache_read": 0.4
1530
- }
1501
+ "input": 0.75,
1502
+ "output": 3.75,
1503
+ "cache_read": 0.075,
1504
+ "input_audio": 0.75
1531
1505
  }
1532
1506
  },
1533
- "gemini-3.1-flash-live-preview": {
1534
- "id": "gemini-3.1-flash-live-preview",
1535
- "name": "Gemini 3.1 Flash Live Preview",
1536
- "description": "High-quality, low-latency Live API model for real-time dialogue and voice-first AI applications",
1507
+ "gemini-3.1-flash-image-preview": {
1508
+ "id": "gemini-3.1-flash-image-preview",
1509
+ "name": "Nano Banana 2",
1510
+ "description": "Image model for prompt-driven generation, editing, and visual design workflows",
1537
1511
  "family": "gemini-flash",
1538
1512
  "attachment": true,
1539
1513
  "reasoning": true,
@@ -1542,28 +1516,56 @@
1542
1516
  "type": "effort",
1543
1517
  "values": [
1544
1518
  "minimal",
1545
- "low",
1546
- "medium",
1547
1519
  "high"
1548
1520
  ]
1549
1521
  }
1550
1522
  ],
1551
- "tool_call": true,
1552
- "structured_output": false,
1523
+ "tool_call": false,
1553
1524
  "temperature": true,
1554
1525
  "knowledge": "2025-01",
1555
- "release_date": "2026-03-26",
1556
- "last_updated": "2026-03-26",
1526
+ "release_date": "2026-02-26",
1527
+ "last_updated": "2026-02-26",
1557
1528
  "modalities": {
1558
1529
  "input": [
1559
1530
  "text",
1560
1531
  "image",
1561
- "video",
1562
- "audio"
1532
+ "pdf"
1563
1533
  ],
1564
1534
  "output": [
1565
1535
  "text",
1566
- "audio"
1536
+ "image"
1537
+ ]
1538
+ },
1539
+ "open_weights": false,
1540
+ "limit": {
1541
+ "context": 65536,
1542
+ "output": 65536
1543
+ },
1544
+ "cost": {
1545
+ "input": 0.5,
1546
+ "output": 60
1547
+ }
1548
+ },
1549
+ "gemini-omni-flash-preview": {
1550
+ "id": "gemini-omni-flash-preview",
1551
+ "name": "Gemini Omni Flash Preview",
1552
+ "description": "Video generation and editing model for fast, conversational text- and image-to-video workflows",
1553
+ "family": "gemini",
1554
+ "attachment": true,
1555
+ "reasoning": true,
1556
+ "reasoning_options": [],
1557
+ "tool_call": false,
1558
+ "temperature": true,
1559
+ "release_date": "2026-06-30",
1560
+ "last_updated": "2026-06-30",
1561
+ "modalities": {
1562
+ "input": [
1563
+ "text",
1564
+ "image",
1565
+ "video"
1566
+ ],
1567
+ "output": [
1568
+ "video"
1567
1569
  ]
1568
1570
  },
1569
1571
  "open_weights": false,
@@ -1572,41 +1574,39 @@
1572
1574
  "output": 65536
1573
1575
  },
1574
1576
  "cost": {
1575
- "input": 0.75,
1576
- "output": 4.5,
1577
- "input_audio": 3,
1578
- "output_audio": 12
1577
+ "input": 1.5,
1578
+ "output": 17.5
1579
1579
  }
1580
1580
  },
1581
- "gemini-3.7-flash": {
1582
- "id": "gemini-3.7-flash",
1583
- "name": "Gemini 3.7 Flash",
1584
- "description": "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning",
1581
+ "gemini-2.5-flash": {
1582
+ "id": "gemini-2.5-flash",
1583
+ "name": "Gemini 2.5 Flash",
1584
+ "description": "Fast Gemini workhorse for multimodal apps where latency and price matter",
1585
1585
  "family": "gemini-flash",
1586
1586
  "attachment": true,
1587
1587
  "reasoning": true,
1588
1588
  "reasoning_options": [
1589
1589
  {
1590
- "type": "effort",
1591
- "values": [
1592
- "low",
1593
- "medium",
1594
- "high"
1595
- ]
1590
+ "type": "toggle"
1591
+ },
1592
+ {
1593
+ "type": "budget_tokens",
1594
+ "min": 0,
1595
+ "max": 24576
1596
1596
  }
1597
1597
  ],
1598
1598
  "tool_call": true,
1599
1599
  "structured_output": true,
1600
1600
  "temperature": true,
1601
- "knowledge": "2026-03",
1602
- "release_date": "2026-08-13",
1603
- "last_updated": "2026-08-13",
1601
+ "knowledge": "2025-01",
1602
+ "release_date": "2025-06-17",
1603
+ "last_updated": "2025-06-17",
1604
1604
  "modalities": {
1605
1605
  "input": [
1606
1606
  "text",
1607
1607
  "image",
1608
- "video",
1609
1608
  "audio",
1609
+ "video",
1610
1610
  "pdf"
1611
1611
  ],
1612
1612
  "output": [
@@ -1619,10 +1619,10 @@
1619
1619
  "output": 65536
1620
1620
  },
1621
1621
  "cost": {
1622
- "input": 0.75,
1623
- "output": 3.75,
1624
- "cache_read": 0.075,
1625
- "input_audio": 0.75
1622
+ "input": 0.3,
1623
+ "output": 2.5,
1624
+ "cache_read": 0.03,
1625
+ "input_audio": 1
1626
1626
  }
1627
1627
  }
1628
1628
  }