llm.rb 15.1.0 → 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 (87) hide show
  1. checksums.yaml +4 -4
  2. data/CHANGELOG.md +296 -11
  3. data/README.md +68 -43
  4. data/bin/llm.rb +12 -4
  5. data/data/alibaba.json +912 -823
  6. data/data/anthropic.json +234 -187
  7. data/data/bedrock.json +3611 -2217
  8. data/data/deepinfra.json +1164 -989
  9. data/data/deepseek.json +65 -81
  10. data/data/google.json +668 -666
  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 +7798 -7661
  15. data/data/xai.json +194 -194
  16. data/data/zai.json +242 -149
  17. data/docs/deepdive/advanced/compaction.md +5 -5
  18. data/docs/deepdive/advanced/guard.md +2 -2
  19. data/docs/deepdive/features/builtin_tools.md +93 -22
  20. data/docs/deepdive/features/{repl.md → console.md} +28 -28
  21. data/docs/deepdive/features/database.md +3 -3
  22. data/docs/deepdive/fundamentals/agents.md +2 -2
  23. data/docs/deepdive/fundamentals/providers.md +91 -6
  24. data/docs/deepdive/fundamentals/skills.md +14 -6
  25. data/docs/deepdive/fundamentals/tools.md +63 -31
  26. data/docs/deepdive/reference/cost.md +2 -2
  27. data/docs/deepdive/reference/model_registry.md +2 -2
  28. data/docs/deepdive/reference/tracer.md +15 -13
  29. data/docs/deepdive.md +2 -2
  30. data/lib/llm/active_record/acts_as_agent.rb +9 -5
  31. data/lib/llm/agent.rb +34 -15
  32. data/lib/llm/{repl → console}/bar.rb +3 -3
  33. data/lib/llm/{repl → console}/buffer.rb +4 -4
  34. data/lib/llm/{repl → console}/color.rb +2 -2
  35. data/lib/llm/{repl → console}/command.rb +12 -12
  36. data/lib/llm/{repl → console}/commands/exit.rb +4 -4
  37. data/lib/llm/{repl → console}/commands/help.rb +1 -1
  38. data/lib/llm/{repl/commands/compact.rb → console/commands/keep.rb} +11 -9
  39. data/lib/llm/{repl → console}/commands/model.rb +2 -2
  40. data/lib/llm/{repl → console}/input/cache.rb +2 -2
  41. data/lib/llm/{repl → console}/input/char.rb +2 -2
  42. data/lib/llm/{repl → console}/input/row.rb +1 -1
  43. data/lib/llm/{repl → console}/input.rb +13 -6
  44. data/lib/llm/console/markdown/parser.rb +78 -0
  45. data/lib/llm/{repl → console}/markdown/table.rb +3 -3
  46. data/lib/llm/{repl → console}/markdown.rb +13 -30
  47. data/lib/llm/{repl → console}/node.rb +3 -3
  48. data/lib/llm/{repl → console}/status.rb +11 -11
  49. data/lib/llm/{repl → console}/stream.rb +9 -9
  50. data/lib/llm/{repl → console}/walker.rb +1 -1
  51. data/lib/llm/{repl → console}/window.rb +10 -10
  52. data/lib/llm/{repl.rb → console.rb} +38 -19
  53. data/lib/llm/context/deserializer.rb +2 -1
  54. data/lib/llm/context.rb +1 -0
  55. data/lib/llm/function/async/reactor.rb +20 -1
  56. data/lib/llm/function.rb +1 -1
  57. data/lib/llm/json_adapter.rb +40 -28
  58. data/lib/llm/message.rb +7 -0
  59. data/lib/llm/provider.rb +2 -2
  60. data/lib/llm/providers/alibaba.rb +1 -1
  61. data/lib/llm/providers/deepseek.rb +1 -1
  62. data/lib/llm/providers/openai.rb +1 -0
  63. data/lib/llm/schema/leaf.rb +34 -2
  64. data/lib/llm/schema.rb +4 -2
  65. data/lib/llm/sequel/agent.rb +9 -5
  66. data/lib/llm/tool/param.rb +5 -1
  67. data/lib/llm/tool.rb +5 -0
  68. data/lib/llm/tools/bundle.rb +53 -0
  69. data/lib/llm/tools/edit-file.rb +7 -2
  70. data/lib/llm/tools/exec.rb +78 -0
  71. data/lib/llm/tools/git.rb +27 -26
  72. data/lib/llm/tools/mkdir.rb +12 -19
  73. data/lib/llm/tools/read_file.rb +69 -9
  74. data/lib/llm/tools/rg.rb +20 -24
  75. data/lib/llm/tools/ruby.rb +17 -25
  76. data/lib/llm/tools/utils.rb +74 -1
  77. data/lib/llm/tools/write_file.rb +4 -1
  78. data/lib/llm/tracer/logger.rb +2 -2
  79. data/lib/llm/tracer/pretty_logger.rb +4 -4
  80. data/lib/llm/tracer/telemetry.rb +2 -2
  81. data/lib/llm/tracer.rb +33 -0
  82. data/lib/llm/transport/utils.rb +1 -1
  83. data/lib/llm/version.rb +1 -1
  84. data/lib/llm.rb +4 -13
  85. data/llm.gemspec +7 -8
  86. metadata +64 -37
  87. 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": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
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,
@@ -158,75 +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
- "knowledge": "2026-03",
171
- "release_date": "2026-07-21",
172
- "last_updated": "2026-07-21",
125
+ "knowledge": "2025-01",
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
- "input": 0.3,
192
- "output": 2.5,
193
- "cache_read": 0.03
144
+ "input": 0.25,
145
+ "output": 30
194
146
  }
195
147
  },
196
- "gemini-3.5-flash-lite": {
197
- "id": "gemini-3.5-flash-lite",
198
- "name": "Gemini 3.5 Flash Lite",
199
- "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
200
- "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",
201
153
  "attachment": true,
202
- "reasoning": true,
203
- "reasoning_options": [
204
- {
205
- "type": "effort",
206
- "values": [
207
- "minimal",
208
- "low",
209
- "medium",
210
- "high"
211
- ]
212
- }
213
- ],
214
- "tool_call": true,
215
- "structured_output": true,
154
+ "reasoning": false,
155
+ "tool_call": false,
156
+ "structured_output": false,
216
157
  "temperature": true,
217
- "knowledge": "2026-03",
218
- "release_date": "2026-07-21",
219
- "last_updated": "2026-07-21",
158
+ "release_date": "2026-03-25",
159
+ "last_updated": "2026-03-25",
220
160
  "modalities": {
221
161
  "input": [
222
162
  "text",
223
- "image",
224
- "video",
225
- "audio",
226
- "pdf"
163
+ "image"
227
164
  ],
228
165
  "output": [
229
- "text"
166
+ "text",
167
+ "audio"
230
168
  ]
231
169
  },
232
170
  "open_weights": false,
@@ -235,33 +173,23 @@
235
173
  "output": 65536
236
174
  },
237
175
  "cost": {
238
- "input": 0.3,
239
- "output": 2.5,
240
- "cache_read": 0.03
176
+ "input": 0,
177
+ "output": 0
241
178
  }
242
179
  },
243
- "gemini-3.1-flash-lite-image": {
244
- "id": "gemini-3.1-flash-lite-image",
245
- "name": "Nano Banana 2 Lite",
246
- "description": "Fastest, most cost-efficient Gemini image model for high-volume 1K generation and editing",
247
- "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",
248
185
  "attachment": true,
249
186
  "reasoning": true,
250
- "reasoning_options": [
251
- {
252
- "type": "effort",
253
- "values": [
254
- "minimal",
255
- "high"
256
- ]
257
- }
258
- ],
259
- "tool_call": true,
260
- "structured_output": false,
187
+ "reasoning_options": [],
188
+ "tool_call": false,
261
189
  "temperature": true,
262
- "knowledge": "2025-01",
263
- "release_date": "2026-06-30",
264
- "last_updated": "2026-06-30",
190
+ "knowledge": "2024-06",
191
+ "release_date": "2025-08-26",
192
+ "last_updated": "2025-08-26",
265
193
  "modalities": {
266
194
  "input": [
267
195
  "text",
@@ -274,37 +202,28 @@
274
202
  },
275
203
  "open_weights": false,
276
204
  "limit": {
277
- "context": 65536,
278
- "output": 65536
205
+ "context": 32768,
206
+ "output": 32768
279
207
  },
280
208
  "cost": {
281
- "input": 0.25,
282
- "output": 30
209
+ "input": 0.3,
210
+ "output": 30,
211
+ "cache_read": 0.075
283
212
  }
284
213
  },
285
- "gemini-3.1-pro-preview": {
286
- "id": "gemini-3.1-pro-preview",
287
- "name": "Gemini 3.1 Pro Preview",
288
- "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",
289
218
  "family": "gemini-pro",
290
219
  "attachment": true,
291
220
  "reasoning": true,
292
- "reasoning_options": [
293
- {
294
- "type": "effort",
295
- "values": [
296
- "low",
297
- "medium",
298
- "high"
299
- ]
300
- }
301
- ],
221
+ "reasoning_options": [],
302
222
  "tool_call": true,
303
- "structured_output": true,
304
223
  "temperature": true,
305
224
  "knowledge": "2025-01",
306
- "release_date": "2026-02-19",
307
- "last_updated": "2026-02-19",
225
+ "release_date": "2026-04-21",
226
+ "last_updated": "2026-04-21",
308
227
  "modalities": {
309
228
  "input": [
310
229
  "text",
@@ -314,12 +233,13 @@
314
233
  "pdf"
315
234
  ],
316
235
  "output": [
317
- "text"
236
+ "text",
237
+ "image"
318
238
  ]
319
239
  },
320
240
  "open_weights": false,
321
241
  "limit": {
322
- "context": 1048576,
242
+ "context": 131072,
323
243
  "output": 65536
324
244
  },
325
245
  "cost": {
@@ -344,10 +264,51 @@
344
264
  }
345
265
  }
346
266
  },
347
- "gemini-3.1-pro-preview-customtools": {
348
- "id": "gemini-3.1-pro-preview-customtools",
349
- "name": "Gemini 3.1 Pro Preview Custom Tools",
350
- "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",
351
312
  "family": "gemini-pro",
352
313
  "attachment": true,
353
314
  "reasoning": true,
@@ -506,34 +467,83 @@
506
467
  "input_audio": 0.3
507
468
  }
508
469
  },
509
- "gemini-robotics-er-1.6-preview": {
510
- "id": "gemini-robotics-er-1.6-preview",
511
- "name": "Gemini Robotics-ER 1.6 Preview",
512
- "description": "Vision-language model for embodied reasoning: spatial understanding, task planning, and physical-world agentic robotics",
513
- "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",
514
521
  "attachment": true,
515
522
  "reasoning": true,
516
523
  "reasoning_options": [
517
524
  {
518
- "type": "toggle"
519
- },
520
- {
521
- "type": "budget_tokens",
522
- "min": 0
525
+ "type": "effort",
526
+ "values": [
527
+ "minimal",
528
+ "low",
529
+ "medium",
530
+ "high"
531
+ ]
523
532
  }
524
533
  ],
525
534
  "tool_call": true,
526
535
  "structured_output": true,
527
536
  "temperature": true,
528
- "knowledge": "2025-01",
529
- "release_date": "2026-04-14",
530
- "last_updated": "2026-04-14",
537
+ "knowledge": "2026-03",
538
+ "release_date": "2026-07-21",
539
+ "last_updated": "2026-07-21",
531
540
  "modalities": {
532
541
  "input": [
533
542
  "text",
534
543
  "image",
535
544
  "video",
536
- "audio"
545
+ "audio",
546
+ "pdf"
537
547
  ],
538
548
  "output": [
539
549
  "text"
@@ -541,67 +551,98 @@
541
551
  },
542
552
  "open_weights": false,
543
553
  "limit": {
544
- "context": 131072,
554
+ "context": 1048576,
545
555
  "output": 65536
546
556
  },
547
557
  "cost": {
548
- "input": 1,
549
- "output": 5,
550
- "input_audio": 2
558
+ "input": 0.75,
559
+ "output": 3.75,
560
+ "cache_read": 0.075,
561
+ "input_audio": 0.75
551
562
  }
552
563
  },
553
- "gemma-4-31b-it": {
554
- "id": "gemma-4-31b-it",
555
- "name": "Gemma 4 31B IT",
556
- "description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
557
- "family": "gemma",
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",
558
569
  "attachment": true,
559
570
  "reasoning": true,
560
571
  "reasoning_options": [
561
572
  {
562
- "type": "toggle"
573
+ "type": "effort",
574
+ "values": [
575
+ "minimal",
576
+ "low",
577
+ "medium",
578
+ "high"
579
+ ]
563
580
  }
564
581
  ],
565
582
  "tool_call": true,
566
583
  "structured_output": true,
567
584
  "temperature": true,
568
- "release_date": "2026-04-02",
569
- "last_updated": "2026-04-02",
585
+ "knowledge": "2025-01",
586
+ "release_date": "2026-05-07",
587
+ "last_updated": "2026-05-07",
570
588
  "modalities": {
571
589
  "input": [
572
590
  "text",
573
- "image"
591
+ "image",
592
+ "video",
593
+ "audio",
594
+ "pdf"
574
595
  ],
575
596
  "output": [
576
597
  "text"
577
598
  ]
578
599
  },
579
- "open_weights": true,
600
+ "open_weights": false,
580
601
  "limit": {
581
- "context": 262144,
582
- "output": 32768
602
+ "context": 1048576,
603
+ "output": 65536
604
+ },
605
+ "cost": {
606
+ "input": 0.25,
607
+ "output": 1.5,
608
+ "cache_read": 0.025,
609
+ "input_audio": 0.5
583
610
  }
584
611
  },
585
- "gemini-2.5-computer-use-preview-10-2025": {
586
- "id": "gemini-2.5-computer-use-preview-10-2025",
587
- "name": "Gemini 2.5 Computer Use Preview 10-2025",
588
- "description": "Specialized Gemini 2.5 model for browser-control agents that automate UI tasks",
589
- "family": "gemini-pro",
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",
590
617
  "attachment": true,
591
618
  "reasoning": true,
592
- "reasoning_options": [],
619
+ "reasoning_options": [
620
+ {
621
+ "type": "effort",
622
+ "values": [
623
+ "minimal",
624
+ "low",
625
+ "medium",
626
+ "high"
627
+ ]
628
+ }
629
+ ],
593
630
  "tool_call": true,
631
+ "structured_output": false,
594
632
  "temperature": true,
595
633
  "knowledge": "2025-01",
596
- "release_date": "2025-10-07",
597
- "last_updated": "2025-10-07",
634
+ "release_date": "2026-03-26",
635
+ "last_updated": "2026-03-26",
598
636
  "modalities": {
599
637
  "input": [
600
638
  "text",
601
- "image"
639
+ "image",
640
+ "video",
641
+ "audio"
602
642
  ],
603
643
  "output": [
604
- "text"
644
+ "text",
645
+ "audio"
605
646
  ]
606
647
  },
607
648
  "open_weights": false,
@@ -610,44 +651,151 @@
610
651
  "output": 65536
611
652
  },
612
653
  "cost": {
613
- "input": 1.25,
614
- "output": 10,
615
- "tiers": [
616
- {
617
- "input": 2.5,
618
- "output": 15,
619
- "tier": {
620
- "type": "context",
621
- "size": 200000
622
- }
623
- }
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": [
674
+ "text"
624
675
  ],
625
- "context_over_200k": {
626
- "input": 2.5,
627
- "output": 15
628
- }
676
+ "output": [
677
+ "audio"
678
+ ]
679
+ },
680
+ "open_weights": false,
681
+ "limit": {
682
+ "context": 8192,
683
+ "output": 16384
684
+ },
685
+ "cost": {
686
+ "input": 1,
687
+ "output": 20
629
688
  }
630
689
  },
631
- "lyria-3-clip-preview": {
632
- "id": "lyria-3-clip-preview",
633
- "name": "Lyria 3 Clip Preview",
634
- "description": "Music generation model for short 30-second clips, loops, and previews from text or image prompts",
635
- "family": "lyria",
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",
695
+ "attachment": true,
696
+ "reasoning": true,
697
+ "reasoning_options": [
698
+ {
699
+ "type": "effort",
700
+ "values": [
701
+ "minimal",
702
+ "low",
703
+ "medium",
704
+ "high"
705
+ ]
706
+ }
707
+ ],
708
+ "tool_call": true,
709
+ "structured_output": true,
710
+ "temperature": true,
711
+ "knowledge": "2025-01",
712
+ "release_date": "2026-05-19",
713
+ "last_updated": "2026-05-19",
714
+ "modalities": {
715
+ "input": [
716
+ "text",
717
+ "image",
718
+ "video",
719
+ "audio",
720
+ "pdf"
721
+ ],
722
+ "output": [
723
+ "text"
724
+ ]
725
+ },
726
+ "open_weights": false,
727
+ "limit": {
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",
636
743
  "attachment": true,
637
744
  "reasoning": false,
638
745
  "tool_call": false,
639
- "structured_output": false,
640
- "temperature": true,
641
- "release_date": "2026-03-25",
642
- "last_updated": "2026-03-25",
746
+ "temperature": false,
747
+ "release_date": "2025-10-15",
748
+ "last_updated": "2026-01",
643
749
  "modalities": {
644
750
  "input": [
645
751
  "text",
646
752
  "image"
647
753
  ],
648
754
  "output": [
755
+ "video"
756
+ ]
757
+ },
758
+ "open_weights": false,
759
+ "limit": {
760
+ "context": 480,
761
+ "output": 8192
762
+ },
763
+ "status": "beta"
764
+ },
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",
770
+ "attachment": true,
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,
785
+ "temperature": true,
786
+ "knowledge": "2025-01",
787
+ "release_date": "2026-03-03",
788
+ "last_updated": "2026-03-03",
789
+ "modalities": {
790
+ "input": [
649
791
  "text",
650
- "audio"
792
+ "image",
793
+ "video",
794
+ "audio",
795
+ "pdf"
796
+ ],
797
+ "output": [
798
+ "text"
651
799
  ]
652
800
  },
653
801
  "open_weights": false,
@@ -655,9 +803,12 @@
655
803
  "context": 1048576,
656
804
  "output": 65536
657
805
  },
806
+ "status": "deprecated",
658
807
  "cost": {
659
- "input": 0,
660
- "output": 0
808
+ "input": 0.25,
809
+ "output": 1.5,
810
+ "cache_read": 0.025,
811
+ "input_audio": 0.5
661
812
  }
662
813
  },
663
814
  "veo-3.1-fast-generate-preview": {
@@ -687,70 +838,70 @@
687
838
  "output": 8192
688
839
  }
689
840
  },
690
- "gemini-embedding-001": {
691
- "id": "gemini-embedding-001",
692
- "name": "Gemini Embedding 001",
693
- "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
694
- "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",
695
846
  "attachment": false,
696
847
  "reasoning": false,
697
848
  "tool_call": false,
698
- "temperature": false,
699
- "knowledge": "2025-05",
700
- "release_date": "2025-05-20",
701
- "last_updated": "2025-05-20",
849
+ "temperature": true,
850
+ "knowledge": "2025-01",
851
+ "release_date": "2025-05-01",
852
+ "last_updated": "2025-05-01",
702
853
  "modalities": {
703
854
  "input": [
704
855
  "text"
705
856
  ],
706
857
  "output": [
707
- "text"
858
+ "audio"
708
859
  ]
709
860
  },
710
861
  "open_weights": false,
711
862
  "limit": {
712
- "context": 2048,
713
- "output": 1
863
+ "context": 8192,
864
+ "output": 16384
714
865
  },
715
866
  "cost": {
716
- "input": 0.15,
717
- "output": 0
867
+ "input": 0.5,
868
+ "output": 10
718
869
  }
719
870
  },
720
- "gemini-2.5-pro-preview-tts": {
721
- "id": "gemini-2.5-pro-preview-tts",
722
- "name": "Gemini 2.5 Pro Preview TTS",
723
- "description": "Speech generation model for controllable voice, narration, and audio delivery",
724
- "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",
725
876
  "attachment": false,
726
877
  "reasoning": false,
727
878
  "tool_call": false,
728
- "temperature": true,
729
- "knowledge": "2025-01",
730
- "release_date": "2025-05-01",
731
- "last_updated": "2025-05-01",
879
+ "temperature": false,
880
+ "knowledge": "2025-05",
881
+ "release_date": "2025-05-20",
882
+ "last_updated": "2025-05-20",
732
883
  "modalities": {
733
884
  "input": [
734
885
  "text"
735
886
  ],
736
887
  "output": [
737
- "audio"
888
+ "text"
738
889
  ]
739
890
  },
740
891
  "open_weights": false,
741
892
  "limit": {
742
- "context": 8192,
743
- "output": 16384
893
+ "context": 2048,
894
+ "output": 1
744
895
  },
745
896
  "cost": {
746
- "input": 1,
747
- "output": 20
897
+ "input": 0.15,
898
+ "output": 0
748
899
  }
749
900
  },
750
- "gemini-flash-latest": {
751
- "id": "gemini-flash-latest",
752
- "name": "Gemini Flash Latest",
753
- "description": "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning",
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",
754
905
  "family": "gemini-flash",
755
906
  "attachment": true,
756
907
  "reasoning": true,
@@ -758,47 +909,43 @@
758
909
  {
759
910
  "type": "effort",
760
911
  "values": [
761
- "low",
762
- "medium",
912
+ "minimal",
763
913
  "high"
764
914
  ]
765
915
  }
766
916
  ],
767
- "tool_call": true,
768
- "structured_output": true,
917
+ "tool_call": false,
769
918
  "temperature": true,
770
- "knowledge": "2026-03",
771
- "release_date": "2026-08-13",
772
- "last_updated": "2026-08-13",
919
+ "knowledge": "2025-01",
920
+ "release_date": "2026-05-28",
921
+ "last_updated": "2026-05-28",
773
922
  "modalities": {
774
923
  "input": [
775
924
  "text",
776
925
  "image",
777
926
  "video",
778
- "audio",
779
927
  "pdf"
780
928
  ],
781
929
  "output": [
782
- "text"
930
+ "text",
931
+ "image"
783
932
  ]
784
933
  },
785
934
  "open_weights": false,
786
935
  "limit": {
787
- "context": 1048576,
936
+ "context": 65536,
788
937
  "output": 65536
789
938
  },
790
939
  "cost": {
791
- "input": 0.75,
792
- "output": 3.75,
793
- "cache_read": 0.075,
794
- "input_audio": 0.75
940
+ "input": 0.5,
941
+ "output": 60
795
942
  }
796
943
  },
797
- "gemini-3.5-flash": {
798
- "id": "gemini-3.5-flash",
799
- "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",
800
947
  "description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
801
- "family": "gemini-flash",
948
+ "family": "gemini-flash-lite",
802
949
  "attachment": true,
803
950
  "reasoning": true,
804
951
  "reasoning_options": [
@@ -815,9 +962,9 @@
815
962
  "tool_call": true,
816
963
  "structured_output": true,
817
964
  "temperature": true,
818
- "knowledge": "2025-01",
819
- "release_date": "2026-05-19",
820
- "last_updated": "2026-05-19",
965
+ "knowledge": "2026-03",
966
+ "release_date": "2026-07-21",
967
+ "last_updated": "2026-07-21",
821
968
  "modalities": {
822
969
  "input": [
823
970
  "text",
@@ -836,88 +983,90 @@
836
983
  "output": 65536
837
984
  },
838
985
  "cost": {
839
- "input": 1.5,
840
- "output": 9,
841
- "cache_read": 0.15,
842
- "input_audio": 1.5
986
+ "input": 0.3,
987
+ "output": 2.5,
988
+ "cache_read": 0.03
843
989
  }
844
990
  },
845
- "gemini-3.5-live-translate-preview": {
846
- "id": "gemini-3.5-live-translate-preview",
847
- "name": "Gemini 3.5 Live Translate Preview",
848
- "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",
849
995
  "family": "gemini-pro",
850
- "attachment": false,
851
- "reasoning": false,
996
+ "attachment": true,
997
+ "reasoning": true,
998
+ "reasoning_options": [],
852
999
  "tool_call": false,
853
1000
  "temperature": true,
854
1001
  "knowledge": "2025-01",
855
- "release_date": "2026-06-09",
856
- "last_updated": "2026-06-09",
1002
+ "release_date": "2025-11-20",
1003
+ "last_updated": "2025-11-20",
857
1004
  "modalities": {
858
1005
  "input": [
859
- "audio"
1006
+ "text",
1007
+ "image"
860
1008
  ],
861
1009
  "output": [
862
- "audio",
863
- "text"
1010
+ "text",
1011
+ "image"
864
1012
  ]
865
1013
  },
866
1014
  "open_weights": false,
867
1015
  "limit": {
868
- "context": 16384,
1016
+ "context": 131072,
869
1017
  "output": 32768
870
1018
  },
871
1019
  "cost": {
872
- "input": 3.5,
873
- "output": 21,
874
- "input_audio": 3.5,
875
- "output_audio": 21
1020
+ "input": 2,
1021
+ "output": 120
876
1022
  }
877
1023
  },
878
- "veo-3.1-lite-generate-preview": {
879
- "id": "veo-3.1-lite-generate-preview",
880
- "name": "Veo 3.1 lite",
881
- "description": "Video model for prompt-guided generation, editing, and motion workflows",
882
- "family": "veo",
883
- "attachment": true,
884
- "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": [],
885
1032
  "tool_call": false,
886
- "temperature": false,
887
- "release_date": "2026-03-31",
888
- "last_updated": "2026-03-31",
1033
+ "temperature": true,
1034
+ "knowledge": "2025-01",
1035
+ "release_date": "2026-04-15",
1036
+ "last_updated": "2026-04-15",
889
1037
  "modalities": {
890
1038
  "input": [
891
- "text",
892
- "image"
1039
+ "text"
893
1040
  ],
894
1041
  "output": [
895
- "video"
1042
+ "audio"
896
1043
  ]
897
1044
  },
898
1045
  "open_weights": false,
899
1046
  "limit": {
900
- "context": 480,
901
- "output": 8192
1047
+ "context": 8192,
1048
+ "output": 16384
1049
+ },
1050
+ "cost": {
1051
+ "input": 1,
1052
+ "output": 20
902
1053
  }
903
1054
  },
904
- "gemini-omni-flash-preview": {
905
- "id": "gemini-omni-flash-preview",
906
- "name": "Gemini Omni Flash Preview",
907
- "description": "Video generation and editing model for fast, conversational text- and image-to-video workflows",
908
- "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",
909
1060
  "attachment": true,
910
- "reasoning": true,
911
- "reasoning_options": [],
1061
+ "reasoning": false,
912
1062
  "tool_call": false,
913
- "temperature": true,
914
- "release_date": "2026-06-30",
915
- "last_updated": "2026-06-30",
1063
+ "temperature": false,
1064
+ "release_date": "2026-03-31",
1065
+ "last_updated": "2026-03-31",
916
1066
  "modalities": {
917
1067
  "input": [
918
1068
  "text",
919
- "image",
920
- "video"
1069
+ "image"
921
1070
  ],
922
1071
  "output": [
923
1072
  "video"
@@ -925,18 +1074,14 @@
925
1074
  },
926
1075
  "open_weights": false,
927
1076
  "limit": {
928
- "context": 131072,
929
- "output": 65536
930
- },
931
- "cost": {
932
- "input": 1.5,
933
- "output": 17.5
1077
+ "context": 480,
1078
+ "output": 8192
934
1079
  }
935
1080
  },
936
- "gemini-3.1-flash-lite": {
937
- "id": "gemini-3.1-flash-lite",
938
- "name": "Gemini 3.1 Flash Lite",
939
- "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",
940
1085
  "family": "gemini-flash-lite",
941
1086
  "attachment": true,
942
1087
  "reasoning": true,
@@ -954,9 +1099,9 @@
954
1099
  "tool_call": true,
955
1100
  "structured_output": true,
956
1101
  "temperature": true,
957
- "knowledge": "2025-01",
958
- "release_date": "2026-05-07",
959
- "last_updated": "2026-05-07",
1102
+ "knowledge": "2026-03",
1103
+ "release_date": "2026-07-21",
1104
+ "last_updated": "2026-07-21",
960
1105
  "modalities": {
961
1106
  "input": [
962
1107
  "text",
@@ -975,116 +1120,43 @@
975
1120
  "output": 65536
976
1121
  },
977
1122
  "cost": {
978
- "input": 0.25,
979
- "output": 1.5,
980
- "cache_read": 0.025,
981
- "input_audio": 0.5
1123
+ "input": 0.3,
1124
+ "output": 2.5,
1125
+ "cache_read": 0.03
982
1126
  }
983
1127
  },
984
- "gemini-2.5-flash": {
985
- "id": "gemini-2.5-flash",
986
- "name": "Gemini 2.5 Flash",
987
- "description": "Fast Gemini workhorse for multimodal apps where latency and price matter",
988
- "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",
989
1133
  "attachment": true,
990
1134
  "reasoning": true,
991
1135
  "reasoning_options": [
992
1136
  {
993
1137
  "type": "toggle"
994
- },
995
- {
996
- "type": "budget_tokens",
997
- "min": 0,
998
- "max": 24576
999
1138
  }
1000
1139
  ],
1001
1140
  "tool_call": true,
1002
1141
  "structured_output": true,
1003
1142
  "temperature": true,
1004
- "knowledge": "2025-01",
1005
- "release_date": "2025-06-17",
1006
- "last_updated": "2025-06-17",
1143
+ "release_date": "2026-04-02",
1144
+ "last_updated": "2026-04-02",
1007
1145
  "modalities": {
1008
1146
  "input": [
1009
1147
  "text",
1010
- "image",
1011
- "audio",
1012
- "video",
1013
- "pdf"
1148
+ "image"
1014
1149
  ],
1015
1150
  "output": [
1016
1151
  "text"
1017
1152
  ]
1018
1153
  },
1019
- "open_weights": false,
1020
- "limit": {
1021
- "context": 1048576,
1022
- "output": 65536
1023
- },
1024
- "cost": {
1025
- "input": 0.3,
1026
- "output": 2.5,
1027
- "cache_read": 0.03,
1028
- "input_audio": 1
1029
- }
1030
- },
1031
- "gemini-2.5-flash-preview-tts": {
1032
- "id": "gemini-2.5-flash-preview-tts",
1033
- "name": "Gemini 2.5 Flash Preview TTS",
1034
- "description": "Speech generation model for controllable voice, narration, and audio delivery",
1035
- "family": "gemini-flash",
1036
- "attachment": false,
1037
- "reasoning": false,
1038
- "tool_call": false,
1039
- "temperature": true,
1040
- "knowledge": "2025-01",
1041
- "release_date": "2025-05-01",
1042
- "last_updated": "2025-05-01",
1043
- "modalities": {
1044
- "input": [
1045
- "text"
1046
- ],
1047
- "output": [
1048
- "audio"
1049
- ]
1050
- },
1051
- "open_weights": false,
1154
+ "open_weights": true,
1052
1155
  "limit": {
1053
- "context": 8192,
1054
- "output": 16384
1055
- },
1056
- "cost": {
1057
- "input": 0.5,
1058
- "output": 10
1156
+ "context": 262144,
1157
+ "output": 32768
1059
1158
  }
1060
1159
  },
1061
- "veo-3.1-generate-preview": {
1062
- "id": "veo-3.1-generate-preview",
1063
- "name": "Veo 3.1",
1064
- "description": "Video model for prompt-guided generation, editing, and motion workflows",
1065
- "family": "veo",
1066
- "attachment": true,
1067
- "reasoning": false,
1068
- "tool_call": false,
1069
- "temperature": false,
1070
- "release_date": "2025-10-15",
1071
- "last_updated": "2026-01",
1072
- "modalities": {
1073
- "input": [
1074
- "text",
1075
- "image"
1076
- ],
1077
- "output": [
1078
- "video"
1079
- ]
1080
- },
1081
- "open_weights": false,
1082
- "limit": {
1083
- "context": 480,
1084
- "output": 8192
1085
- },
1086
- "status": "beta"
1087
- },
1088
1160
  "gemini-embedding-2": {
1089
1161
  "id": "gemini-embedding-2",
1090
1162
  "name": "Gemini Embedding 2",
@@ -1120,10 +1192,10 @@
1120
1192
  "input_audio": 6.5
1121
1193
  }
1122
1194
  },
1123
- "gemini-3.6-flash": {
1124
- "id": "gemini-3.6-flash",
1125
- "name": "Gemini 3.6 Flash",
1126
- "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",
1127
1199
  "family": "gemini-flash",
1128
1200
  "attachment": true,
1129
1201
  "reasoning": true,
@@ -1141,9 +1213,9 @@
1141
1213
  "tool_call": true,
1142
1214
  "structured_output": true,
1143
1215
  "temperature": true,
1144
- "knowledge": "2026-03",
1145
- "release_date": "2026-07-21",
1146
- "last_updated": "2026-07-21",
1216
+ "knowledge": "2025-01",
1217
+ "release_date": "2025-12-17",
1218
+ "last_updated": "2025-12-17",
1147
1219
  "modalities": {
1148
1220
  "input": [
1149
1221
  "text",
@@ -1161,60 +1233,17 @@
1161
1233
  "context": 1048576,
1162
1234
  "output": 65536
1163
1235
  },
1164
- "cost": {
1165
- "input": 0.75,
1166
- "output": 3.75,
1167
- "cache_read": 0.075,
1168
- "input_audio": 0.75
1169
- }
1170
- },
1171
- "gemini-3.1-flash-image": {
1172
- "id": "gemini-3.1-flash-image",
1173
- "name": "Nano Banana 2",
1174
- "description": "Image model for prompt-driven generation, editing, and visual design workflows",
1175
- "family": "gemini-flash",
1176
- "attachment": true,
1177
- "reasoning": true,
1178
- "reasoning_options": [
1179
- {
1180
- "type": "effort",
1181
- "values": [
1182
- "minimal",
1183
- "high"
1184
- ]
1185
- }
1186
- ],
1187
- "tool_call": false,
1188
- "temperature": true,
1189
- "knowledge": "2025-01",
1190
- "release_date": "2026-05-28",
1191
- "last_updated": "2026-05-28",
1192
- "modalities": {
1193
- "input": [
1194
- "text",
1195
- "image",
1196
- "video",
1197
- "pdf"
1198
- ],
1199
- "output": [
1200
- "text",
1201
- "image"
1202
- ]
1203
- },
1204
- "open_weights": false,
1205
- "limit": {
1206
- "context": 65536,
1207
- "output": 65536
1208
- },
1209
1236
  "cost": {
1210
1237
  "input": 0.5,
1211
- "output": 60
1238
+ "output": 3,
1239
+ "cache_read": 0.05,
1240
+ "input_audio": 1
1212
1241
  }
1213
1242
  },
1214
- "gemini-3.1-flash-image-preview": {
1215
- "id": "gemini-3.1-flash-image-preview",
1216
- "name": "Nano Banana 2",
1217
- "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",
1218
1247
  "family": "gemini-flash",
1219
1248
  "attachment": true,
1220
1249
  "reasoning": true,
@@ -1222,103 +1251,94 @@
1222
1251
  {
1223
1252
  "type": "effort",
1224
1253
  "values": [
1225
- "minimal",
1254
+ "low",
1255
+ "medium",
1226
1256
  "high"
1227
1257
  ]
1228
1258
  }
1229
1259
  ],
1230
- "tool_call": false,
1260
+ "tool_call": true,
1261
+ "structured_output": true,
1231
1262
  "temperature": true,
1232
- "knowledge": "2025-01",
1233
- "release_date": "2026-02-26",
1234
- "last_updated": "2026-02-26",
1263
+ "release_date": "2026-09-02",
1264
+ "last_updated": "2026-09-02",
1235
1265
  "modalities": {
1236
1266
  "input": [
1237
1267
  "text",
1238
1268
  "image",
1269
+ "video",
1270
+ "audio",
1239
1271
  "pdf"
1240
1272
  ],
1241
1273
  "output": [
1242
- "text",
1243
- "image"
1274
+ "text"
1244
1275
  ]
1245
1276
  },
1246
1277
  "open_weights": false,
1247
1278
  "limit": {
1248
- "context": 65536,
1279
+ "context": 1048576,
1249
1280
  "output": 65536
1250
1281
  },
1251
1282
  "cost": {
1252
- "input": 0.5,
1253
- "output": 60
1283
+ "input": 0.75,
1284
+ "output": 3.75,
1285
+ "cache_read": 0.075,
1286
+ "input_audio": 0.75
1254
1287
  }
1255
1288
  },
1256
- "lyria-3-pro-preview": {
1257
- "id": "lyria-3-pro-preview",
1258
- "name": "Lyria 3 Pro Preview",
1259
- "description": "Music generation model for full-length songs from text or images with vocals and structure",
1260
- "family": "lyria",
1261
- "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,
1262
1295
  "reasoning": false,
1263
1296
  "tool_call": false,
1264
- "structured_output": false,
1265
1297
  "temperature": true,
1266
- "release_date": "2026-03-25",
1267
- "last_updated": "2026-03-25",
1298
+ "knowledge": "2025-01",
1299
+ "release_date": "2026-06-09",
1300
+ "last_updated": "2026-06-09",
1268
1301
  "modalities": {
1269
1302
  "input": [
1270
- "text",
1271
- "image"
1303
+ "audio"
1272
1304
  ],
1273
1305
  "output": [
1274
- "text",
1275
- "audio"
1306
+ "audio",
1307
+ "text"
1276
1308
  ]
1277
1309
  },
1278
1310
  "open_weights": false,
1279
1311
  "limit": {
1280
- "context": 1048576,
1281
- "output": 65536
1312
+ "context": 16384,
1313
+ "output": 32768
1282
1314
  },
1283
1315
  "cost": {
1284
- "input": 0,
1285
- "output": 0
1316
+ "input": 3.5,
1317
+ "output": 21,
1318
+ "input_audio": 3.5,
1319
+ "output_audio": 21
1286
1320
  }
1287
1321
  },
1288
- "gemini-3.1-flash-lite-preview": {
1289
- "id": "gemini-3.1-flash-lite-preview",
1290
- "name": "Gemini 3.1 Flash Lite Preview",
1291
- "description": "Legacy model retained for compatibility with older integrations",
1292
- "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",
1293
1327
  "attachment": true,
1294
- "reasoning": true,
1295
- "reasoning_options": [
1296
- {
1297
- "type": "effort",
1298
- "values": [
1299
- "minimal",
1300
- "low",
1301
- "medium",
1302
- "high"
1303
- ]
1304
- }
1305
- ],
1306
- "tool_call": true,
1307
- "structured_output": true,
1328
+ "reasoning": false,
1329
+ "tool_call": false,
1330
+ "structured_output": false,
1308
1331
  "temperature": true,
1309
- "knowledge": "2025-01",
1310
- "release_date": "2026-03-03",
1311
- "last_updated": "2026-03-03",
1332
+ "release_date": "2026-03-25",
1333
+ "last_updated": "2026-03-25",
1312
1334
  "modalities": {
1313
1335
  "input": [
1314
1336
  "text",
1315
- "image",
1316
- "video",
1317
- "audio",
1318
- "pdf"
1337
+ "image"
1319
1338
  ],
1320
1339
  "output": [
1321
- "text"
1340
+ "text",
1341
+ "audio"
1322
1342
  ]
1323
1343
  },
1324
1344
  "open_weights": false,
@@ -1326,18 +1346,15 @@
1326
1346
  "context": 1048576,
1327
1347
  "output": 65536
1328
1348
  },
1329
- "status": "deprecated",
1330
1349
  "cost": {
1331
- "input": 0.25,
1332
- "output": 1.5,
1333
- "cache_read": 0.025,
1334
- "input_audio": 0.5
1350
+ "input": 0,
1351
+ "output": 0
1335
1352
  }
1336
1353
  },
1337
- "gemini-3-flash-preview": {
1338
- "id": "gemini-3-flash-preview",
1339
- "name": "Gemini 3 Flash Preview",
1340
- "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",
1341
1358
  "family": "gemini-flash",
1342
1359
  "attachment": true,
1343
1360
  "reasoning": true,
@@ -1345,7 +1362,6 @@
1345
1362
  {
1346
1363
  "type": "effort",
1347
1364
  "values": [
1348
- "minimal",
1349
1365
  "low",
1350
1366
  "medium",
1351
1367
  "high"
@@ -1355,9 +1371,9 @@
1355
1371
  "tool_call": true,
1356
1372
  "structured_output": true,
1357
1373
  "temperature": true,
1358
- "knowledge": "2025-01",
1359
- "release_date": "2025-12-17",
1360
- "last_updated": "2025-12-17",
1374
+ "knowledge": "2026-03",
1375
+ "release_date": "2026-08-13",
1376
+ "last_updated": "2026-08-13",
1361
1377
  "modalities": {
1362
1378
  "input": [
1363
1379
  "text",
@@ -1376,44 +1392,10 @@
1376
1392
  "output": 65536
1377
1393
  },
1378
1394
  "cost": {
1379
- "input": 0.5,
1380
- "output": 3,
1381
- "cache_read": 0.05,
1382
- "input_audio": 1
1383
- }
1384
- },
1385
- "gemini-2.5-flash-image": {
1386
- "id": "gemini-2.5-flash-image",
1387
- "name": "Nano Banana",
1388
- "description": "Nano Banana image model for fast generation, edits, and character-consistent assets",
1389
- "family": "gemini-flash",
1390
- "attachment": true,
1391
- "reasoning": true,
1392
- "reasoning_options": [],
1393
- "tool_call": false,
1394
- "temperature": true,
1395
- "knowledge": "2024-06",
1396
- "release_date": "2025-08-26",
1397
- "last_updated": "2025-08-26",
1398
- "modalities": {
1399
- "input": [
1400
- "text",
1401
- "image"
1402
- ],
1403
- "output": [
1404
- "text",
1405
- "image"
1406
- ]
1407
- },
1408
- "open_weights": false,
1409
- "limit": {
1410
- "context": 32768,
1411
- "output": 32768
1412
- },
1413
- "cost": {
1414
- "input": 0.3,
1415
- "output": 30,
1416
- "cache_read": 0.075
1395
+ "input": 0.75,
1396
+ "output": 3.75,
1397
+ "cache_read": 0.075,
1398
+ "input_audio": 0.75
1417
1399
  }
1418
1400
  },
1419
1401
  "gemini-2.5-pro": {
@@ -1475,19 +1457,29 @@
1475
1457
  }
1476
1458
  }
1477
1459
  },
1478
- "deep-research-max-preview-04-2026": {
1479
- "id": "deep-research-max-preview-04-2026",
1480
- "name": "Deep Research Max Preview (Apr-21-2026)",
1481
- "description": "Maximum-comprehensiveness agentic researcher for multi-step investigation, synthesis, and cited reports",
1482
- "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",
1483
1465
  "attachment": true,
1484
1466
  "reasoning": true,
1485
- "reasoning_options": [],
1467
+ "reasoning_options": [
1468
+ {
1469
+ "type": "effort",
1470
+ "values": [
1471
+ "low",
1472
+ "medium",
1473
+ "high"
1474
+ ]
1475
+ }
1476
+ ],
1486
1477
  "tool_call": true,
1478
+ "structured_output": true,
1487
1479
  "temperature": true,
1488
- "knowledge": "2025-01",
1489
- "release_date": "2026-04-21",
1490
- "last_updated": "2026-04-21",
1480
+ "knowledge": "2026-03",
1481
+ "release_date": "2026-08-13",
1482
+ "last_updated": "2026-08-13",
1491
1483
  "modalities": {
1492
1484
  "input": [
1493
1485
  "text",
@@ -1497,41 +1489,25 @@
1497
1489
  "pdf"
1498
1490
  ],
1499
1491
  "output": [
1500
- "text",
1501
- "image"
1492
+ "text"
1502
1493
  ]
1503
1494
  },
1504
1495
  "open_weights": false,
1505
1496
  "limit": {
1506
- "context": 131072,
1497
+ "context": 1048576,
1507
1498
  "output": 65536
1508
1499
  },
1509
1500
  "cost": {
1510
- "input": 2,
1511
- "output": 12,
1512
- "cache_read": 0.2,
1513
- "tiers": [
1514
- {
1515
- "input": 4,
1516
- "output": 18,
1517
- "cache_read": 0.4,
1518
- "tier": {
1519
- "type": "context",
1520
- "size": 200000
1521
- }
1522
- }
1523
- ],
1524
- "context_over_200k": {
1525
- "input": 4,
1526
- "output": 18,
1527
- "cache_read": 0.4
1528
- }
1501
+ "input": 0.75,
1502
+ "output": 3.75,
1503
+ "cache_read": 0.075,
1504
+ "input_audio": 0.75
1529
1505
  }
1530
1506
  },
1531
- "gemini-3.1-flash-live-preview": {
1532
- "id": "gemini-3.1-flash-live-preview",
1533
- "name": "Gemini 3.1 Flash Live Preview",
1534
- "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",
1535
1511
  "family": "gemini-flash",
1536
1512
  "attachment": true,
1537
1513
  "reasoning": true,
@@ -1540,28 +1516,56 @@
1540
1516
  "type": "effort",
1541
1517
  "values": [
1542
1518
  "minimal",
1543
- "low",
1544
- "medium",
1545
1519
  "high"
1546
1520
  ]
1547
1521
  }
1548
1522
  ],
1549
- "tool_call": true,
1550
- "structured_output": false,
1523
+ "tool_call": false,
1551
1524
  "temperature": true,
1552
1525
  "knowledge": "2025-01",
1553
- "release_date": "2026-03-26",
1554
- "last_updated": "2026-03-26",
1526
+ "release_date": "2026-02-26",
1527
+ "last_updated": "2026-02-26",
1555
1528
  "modalities": {
1556
1529
  "input": [
1557
1530
  "text",
1558
1531
  "image",
1559
- "video",
1560
- "audio"
1532
+ "pdf"
1561
1533
  ],
1562
1534
  "output": [
1563
1535
  "text",
1564
- "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"
1565
1569
  ]
1566
1570
  },
1567
1571
  "open_weights": false,
@@ -1570,41 +1574,39 @@
1570
1574
  "output": 65536
1571
1575
  },
1572
1576
  "cost": {
1573
- "input": 0.75,
1574
- "output": 4.5,
1575
- "input_audio": 3,
1576
- "output_audio": 12
1577
+ "input": 1.5,
1578
+ "output": 17.5
1577
1579
  }
1578
1580
  },
1579
- "gemini-3.7-flash": {
1580
- "id": "gemini-3.7-flash",
1581
- "name": "Gemini 3.7 Flash",
1582
- "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",
1583
1585
  "family": "gemini-flash",
1584
1586
  "attachment": true,
1585
1587
  "reasoning": true,
1586
1588
  "reasoning_options": [
1587
1589
  {
1588
- "type": "effort",
1589
- "values": [
1590
- "low",
1591
- "medium",
1592
- "high"
1593
- ]
1590
+ "type": "toggle"
1591
+ },
1592
+ {
1593
+ "type": "budget_tokens",
1594
+ "min": 0,
1595
+ "max": 24576
1594
1596
  }
1595
1597
  ],
1596
1598
  "tool_call": true,
1597
1599
  "structured_output": true,
1598
1600
  "temperature": true,
1599
- "knowledge": "2026-03",
1600
- "release_date": "2026-08-13",
1601
- "last_updated": "2026-08-13",
1601
+ "knowledge": "2025-01",
1602
+ "release_date": "2025-06-17",
1603
+ "last_updated": "2025-06-17",
1602
1604
  "modalities": {
1603
1605
  "input": [
1604
1606
  "text",
1605
1607
  "image",
1606
- "video",
1607
1608
  "audio",
1609
+ "video",
1608
1610
  "pdf"
1609
1611
  ],
1610
1612
  "output": [
@@ -1617,10 +1619,10 @@
1617
1619
  "output": 65536
1618
1620
  },
1619
1621
  "cost": {
1620
- "input": 0.75,
1621
- "output": 3.75,
1622
- "cache_read": 0.075,
1623
- "input_audio": 0.75
1622
+ "input": 0.3,
1623
+ "output": 2.5,
1624
+ "cache_read": 0.03,
1625
+ "input_audio": 1
1624
1626
  }
1625
1627
  }
1626
1628
  }