llm.rb 12.3.1 → 12.5.0

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 (45) hide show
  1. checksums.yaml +4 -4
  2. data/CHANGELOG.md +395 -0
  3. data/README.md +103 -16
  4. data/data/anthropic.json +249 -249
  5. data/data/bedrock.json +2038 -1879
  6. data/data/deepinfra.json +591 -591
  7. data/data/deepseek.json +31 -31
  8. data/data/google.json +332 -329
  9. data/data/mistral.json +381 -381
  10. data/data/openai.json +1132 -1132
  11. data/data/xai.json +138 -138
  12. data/data/zai.json +165 -165
  13. data/lib/llm/agent.rb +12 -7
  14. data/lib/llm/buffer.rb +15 -0
  15. data/lib/llm/context/deserializer.rb +2 -5
  16. data/lib/llm/context.rb +4 -5
  17. data/lib/llm/function.rb +2 -35
  18. data/lib/llm/message.rb +12 -2
  19. data/lib/llm/provider.rb +11 -1
  20. data/lib/llm/providers/anthropic/stream_parser.rb +3 -12
  21. data/lib/llm/providers/anthropic.rb +11 -0
  22. data/lib/llm/providers/bedrock/stream_parser.rb +4 -13
  23. data/lib/llm/providers/google/stream_parser.rb +3 -12
  24. data/lib/llm/providers/google.rb +7 -0
  25. data/lib/llm/providers/mistral.rb +11 -0
  26. data/lib/llm/providers/ollama/stream_parser.rb +4 -4
  27. data/lib/llm/providers/ollama.rb +11 -0
  28. data/lib/llm/providers/openai/responses/stream_parser.rb +4 -14
  29. data/lib/llm/providers/openai/responses.rb +10 -0
  30. data/lib/llm/providers/openai/stream_parser.rb +5 -15
  31. data/lib/llm/providers/openai.rb +11 -0
  32. data/lib/llm/repl/command.rb +201 -0
  33. data/lib/llm/repl/commands/exit.rb +23 -0
  34. data/lib/llm/repl/commands/help.rb +24 -0
  35. data/lib/llm/repl/input.rb +118 -35
  36. data/lib/llm/repl/stream.rb +43 -8
  37. data/lib/llm/repl/transcript.rb +6 -0
  38. data/lib/llm/repl/window.rb +28 -6
  39. data/lib/llm/repl.rb +125 -27
  40. data/lib/llm/stream.rb +2 -7
  41. data/lib/llm/tools/ls.rb +30 -0
  42. data/lib/llm/tools/which.rb +38 -0
  43. data/lib/llm/version.rb +1 -1
  44. data/resources/deepdive.md +67 -1
  45. metadata +6 -1
data/data/openai.json CHANGED
@@ -7,12 +7,12 @@
7
7
  "name": "OpenAI",
8
8
  "doc": "https://platform.openai.com/docs/models",
9
9
  "models": {
10
- "o3": {
11
- "id": "o3",
12
- "name": "o3",
13
- "description": "Deliberate o-series reasoner for hard math, coding, and multi-step analysis",
14
- "family": "o",
15
- "attachment": true,
10
+ "gpt-5-codex": {
11
+ "id": "gpt-5-codex",
12
+ "name": "GPT-5-Codex",
13
+ "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
14
+ "family": "gpt-codex",
15
+ "attachment": false,
16
16
  "reasoning": true,
17
17
  "reasoning_options": [
18
18
  {
@@ -27,14 +27,13 @@
27
27
  "tool_call": true,
28
28
  "structured_output": true,
29
29
  "temperature": false,
30
- "knowledge": "2024-05",
31
- "release_date": "2025-04-16",
32
- "last_updated": "2025-04-16",
30
+ "knowledge": "2024-09-30",
31
+ "release_date": "2025-09-15",
32
+ "last_updated": "2025-09-15",
33
33
  "modalities": {
34
34
  "input": [
35
35
  "text",
36
- "image",
37
- "pdf"
36
+ "image"
38
37
  ],
39
38
  "output": [
40
39
  "text"
@@ -42,30 +41,44 @@
42
41
  },
43
42
  "open_weights": false,
44
43
  "limit": {
45
- "context": 200000,
46
- "output": 100000
44
+ "context": 400000,
45
+ "input": 272000,
46
+ "output": 128000
47
47
  },
48
48
  "cost": {
49
- "input": 2,
50
- "output": 8,
51
- "cache_read": 0.5
49
+ "input": 1.25,
50
+ "output": 10,
51
+ "cache_read": 0.125
52
52
  }
53
53
  },
54
- "text-embedding-3-large": {
55
- "id": "text-embedding-3-large",
56
- "name": "text-embedding-3-large",
57
- "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
58
- "family": "text-embedding",
59
- "attachment": false,
60
- "reasoning": false,
61
- "tool_call": false,
54
+ "gpt-5.5-pro": {
55
+ "id": "gpt-5.5-pro",
56
+ "name": "GPT-5.5 Pro",
57
+ "description": "Highest-accuracy GPT-5.5 tier for slower, precision-heavy reasoning and coding",
58
+ "family": "gpt-pro",
59
+ "attachment": true,
60
+ "reasoning": true,
61
+ "reasoning_options": [
62
+ {
63
+ "type": "effort",
64
+ "values": [
65
+ "medium",
66
+ "high",
67
+ "xhigh"
68
+ ]
69
+ }
70
+ ],
71
+ "tool_call": true,
72
+ "structured_output": true,
62
73
  "temperature": false,
63
- "knowledge": "2024-01",
64
- "release_date": "2024-01-25",
65
- "last_updated": "2024-01-25",
74
+ "knowledge": "2025-12-01",
75
+ "release_date": "2026-04-23",
76
+ "last_updated": "2026-04-23",
66
77
  "modalities": {
67
78
  "input": [
68
- "text"
79
+ "text",
80
+ "image",
81
+ "pdf"
69
82
  ],
70
83
  "output": [
71
84
  "text"
@@ -73,41 +86,57 @@
73
86
  },
74
87
  "open_weights": false,
75
88
  "limit": {
76
- "context": 8191,
77
- "output": 3072
89
+ "context": 1050000,
90
+ "input": 922000,
91
+ "output": 128000
78
92
  },
79
93
  "cost": {
80
- "input": 0.13,
81
- "output": 0
94
+ "input": 30,
95
+ "output": 180,
96
+ "tiers": [
97
+ {
98
+ "input": 60,
99
+ "output": 270,
100
+ "tier": {
101
+ "type": "context",
102
+ "size": 272000
103
+ }
104
+ }
105
+ ],
106
+ "context_over_200k": {
107
+ "input": 60,
108
+ "output": 270
109
+ }
82
110
  }
83
111
  },
84
- "gpt-5.2-pro": {
85
- "id": "gpt-5.2-pro",
86
- "name": "GPT-5.2 Pro",
87
- "description": "Higher-accuracy GPT-5.2 variant for tougher reasoning and review workflows",
88
- "family": "gpt-pro",
112
+ "o3": {
113
+ "id": "o3",
114
+ "name": "o3",
115
+ "description": "Deliberate o-series reasoner for hard math, coding, and multi-step analysis",
116
+ "family": "o",
89
117
  "attachment": true,
90
118
  "reasoning": true,
91
119
  "reasoning_options": [
92
120
  {
93
121
  "type": "effort",
94
122
  "values": [
123
+ "low",
95
124
  "medium",
96
- "high",
97
- "xhigh"
125
+ "high"
98
126
  ]
99
127
  }
100
128
  ],
101
129
  "tool_call": true,
102
- "structured_output": false,
130
+ "structured_output": true,
103
131
  "temperature": false,
104
- "knowledge": "2025-08-31",
105
- "release_date": "2025-12-11",
106
- "last_updated": "2025-12-11",
132
+ "knowledge": "2024-05",
133
+ "release_date": "2025-04-16",
134
+ "last_updated": "2025-04-16",
107
135
  "modalities": {
108
136
  "input": [
109
137
  "text",
110
- "image"
138
+ "image",
139
+ "pdf"
111
140
  ],
112
141
  "output": [
113
142
  "text"
@@ -115,19 +144,19 @@
115
144
  },
116
145
  "open_weights": false,
117
146
  "limit": {
118
- "context": 400000,
119
- "input": 272000,
120
- "output": 128000
147
+ "context": 200000,
148
+ "output": 100000
121
149
  },
122
150
  "cost": {
123
- "input": 21,
124
- "output": 168
151
+ "input": 2,
152
+ "output": 8,
153
+ "cache_read": 0.5
125
154
  }
126
155
  },
127
- "gpt-5.6": {
128
- "id": "gpt-5.6",
129
- "name": "GPT-5.6",
130
- "description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
156
+ "gpt-5.4": {
157
+ "id": "gpt-5.4",
158
+ "name": "GPT-5.4",
159
+ "description": "Agent-ready GPT for coding and computer-use workflows at a lower cost",
131
160
  "family": "gpt",
132
161
  "attachment": true,
133
162
  "reasoning": true,
@@ -139,17 +168,16 @@
139
168
  "low",
140
169
  "medium",
141
170
  "high",
142
- "xhigh",
143
- "max"
171
+ "xhigh"
144
172
  ]
145
173
  }
146
174
  ],
147
175
  "tool_call": true,
148
176
  "structured_output": true,
149
177
  "temperature": false,
150
- "knowledge": "2026-02-16",
151
- "release_date": "2026-07-09",
152
- "last_updated": "2026-07-09",
178
+ "knowledge": "2025-08-31",
179
+ "release_date": "2026-03-05",
180
+ "last_updated": "2026-03-05",
153
181
  "modalities": {
154
182
  "input": [
155
183
  "text",
@@ -170,39 +198,27 @@
170
198
  "modes": {
171
199
  "fast": {
172
200
  "cost": {
173
- "input": 10,
174
- "output": 60,
175
- "cache_read": 1,
176
- "cache_write": 12.5
201
+ "input": 5,
202
+ "output": 30,
203
+ "cache_read": 0.5
177
204
  },
178
205
  "provider": {
179
206
  "body": {
180
207
  "service_tier": "priority"
181
208
  }
182
209
  }
183
- },
184
- "pro": {
185
- "provider": {
186
- "body": {
187
- "reasoning": {
188
- "mode": "pro"
189
- }
190
- }
191
- }
192
210
  }
193
211
  }
194
212
  },
195
213
  "cost": {
196
- "input": 5,
197
- "output": 30,
198
- "cache_read": 0.5,
199
- "cache_write": 6.25,
214
+ "input": 2.5,
215
+ "output": 15,
216
+ "cache_read": 0.25,
200
217
  "tiers": [
201
218
  {
202
- "input": 10,
203
- "output": 45,
204
- "cache_read": 1,
205
- "cache_write": 12.5,
219
+ "input": 5,
220
+ "output": 22.5,
221
+ "cache_read": 0.5,
206
222
  "tier": {
207
223
  "type": "context",
208
224
  "size": 272000
@@ -210,37 +226,32 @@
210
226
  }
211
227
  ],
212
228
  "context_over_200k": {
213
- "input": 10,
214
- "output": 45,
215
- "cache_read": 1,
216
- "cache_write": 12.5
229
+ "input": 5,
230
+ "output": 22.5,
231
+ "cache_read": 0.5
217
232
  }
218
233
  }
219
234
  },
220
- "gpt-5": {
221
- "id": "gpt-5",
222
- "name": "GPT-5",
223
- "description": "Original GPT-5 workhorse for reasoning, coding, writing, and tool workflows",
224
- "family": "gpt",
235
+ "o4-mini-deep-research": {
236
+ "id": "o4-mini-deep-research",
237
+ "name": "o4-mini-deep-research",
238
+ "description": "Research model for long-horizon investigation, synthesis, and analytical reports",
239
+ "family": "o-mini",
225
240
  "attachment": true,
226
241
  "reasoning": true,
227
242
  "reasoning_options": [
228
243
  {
229
244
  "type": "effort",
230
245
  "values": [
231
- "minimal",
232
- "low",
233
- "medium",
234
- "high"
246
+ "medium"
235
247
  ]
236
248
  }
237
249
  ],
238
250
  "tool_call": true,
239
- "structured_output": true,
240
251
  "temperature": false,
241
- "knowledge": "2024-09-30",
242
- "release_date": "2025-08-07",
243
- "last_updated": "2025-08-07",
252
+ "knowledge": "2024-05",
253
+ "release_date": "2024-06-26",
254
+ "last_updated": "2024-06-26",
244
255
  "modalities": {
245
256
  "input": [
246
257
  "text",
@@ -252,32 +263,42 @@
252
263
  },
253
264
  "open_weights": false,
254
265
  "limit": {
255
- "context": 400000,
256
- "input": 272000,
257
- "output": 128000
266
+ "context": 200000,
267
+ "output": 100000
258
268
  },
259
269
  "cost": {
260
- "input": 1.25,
261
- "output": 10,
262
- "cache_read": 0.125
270
+ "input": 2,
271
+ "output": 8,
272
+ "cache_read": 0.5
263
273
  }
264
274
  },
265
- "gpt-3.5-turbo": {
266
- "id": "gpt-3.5-turbo",
267
- "name": "GPT-3.5-turbo",
268
- "description": "Compact GPT model for low-latency assistance and high-volume workloads",
269
- "family": "gpt",
270
- "attachment": false,
271
- "reasoning": false,
272
- "tool_call": false,
275
+ "gpt-5.4-pro": {
276
+ "id": "gpt-5.4-pro",
277
+ "name": "GPT-5.4 Pro",
278
+ "description": "More exact GPT-5.4 tier for demanding professional reasoning and agent tasks",
279
+ "family": "gpt-pro",
280
+ "attachment": true,
281
+ "reasoning": true,
282
+ "reasoning_options": [
283
+ {
284
+ "type": "effort",
285
+ "values": [
286
+ "medium",
287
+ "high",
288
+ "xhigh"
289
+ ]
290
+ }
291
+ ],
292
+ "tool_call": true,
273
293
  "structured_output": false,
274
- "temperature": true,
275
- "knowledge": "2021-09-01",
276
- "release_date": "2023-03-01",
277
- "last_updated": "2023-11-06",
294
+ "temperature": false,
295
+ "knowledge": "2025-08-31",
296
+ "release_date": "2026-03-05",
297
+ "last_updated": "2026-03-05",
278
298
  "modalities": {
279
299
  "input": [
280
- "text"
300
+ "text",
301
+ "image"
281
302
  ],
282
303
  "output": [
283
304
  "text"
@@ -285,74 +306,130 @@
285
306
  },
286
307
  "open_weights": false,
287
308
  "limit": {
288
- "context": 16385,
289
- "output": 4096
309
+ "context": 1050000,
310
+ "input": 922000,
311
+ "output": 128000
290
312
  },
291
313
  "cost": {
292
- "input": 0.5,
293
- "output": 1.5,
294
- "cache_read": 0
314
+ "input": 30,
315
+ "output": 180,
316
+ "tiers": [
317
+ {
318
+ "input": 60,
319
+ "output": 270,
320
+ "tier": {
321
+ "type": "context",
322
+ "size": 272000
323
+ }
324
+ }
325
+ ],
326
+ "context_over_200k": {
327
+ "input": 60,
328
+ "output": 270
329
+ }
295
330
  }
296
331
  },
297
- "gpt-5-pro": {
298
- "id": "gpt-5-pro",
299
- "name": "GPT-5 Pro",
300
- "description": "Higher-accuracy GPT-5 tier for tough analysis, coding reviews, and planning",
301
- "family": "gpt-pro",
332
+ "gpt-4.1-mini": {
333
+ "id": "gpt-4.1-mini",
334
+ "name": "GPT-4.1 mini",
335
+ "description": "Affordable GPT-4.1 lane for fast coding help and structured extraction",
336
+ "family": "gpt-mini",
337
+ "attachment": true,
338
+ "reasoning": false,
339
+ "tool_call": true,
340
+ "structured_output": true,
341
+ "temperature": true,
342
+ "knowledge": "2024-04",
343
+ "release_date": "2025-04-14",
344
+ "last_updated": "2025-04-14",
345
+ "modalities": {
346
+ "input": [
347
+ "text",
348
+ "image",
349
+ "pdf"
350
+ ],
351
+ "output": [
352
+ "text"
353
+ ]
354
+ },
355
+ "open_weights": false,
356
+ "limit": {
357
+ "context": 1047576,
358
+ "output": 32768
359
+ },
360
+ "cost": {
361
+ "input": 0.4,
362
+ "output": 1.6,
363
+ "cache_read": 0.1
364
+ }
365
+ },
366
+ "gpt-realtime-2.1": {
367
+ "id": "gpt-realtime-2.1",
368
+ "name": "GPT-Realtime-2.1",
369
+ "description": "Realtime speech-to-speech model with configurable reasoning, tool use, and robust voice-agent behavior",
370
+ "family": "gpt",
302
371
  "attachment": true,
303
372
  "reasoning": true,
304
373
  "reasoning_options": [
305
374
  {
306
375
  "type": "effort",
307
376
  "values": [
308
- "high"
377
+ "minimal",
378
+ "low",
379
+ "medium",
380
+ "high",
381
+ "xhigh"
309
382
  ]
310
383
  }
311
384
  ],
312
385
  "tool_call": true,
313
- "structured_output": true,
386
+ "structured_output": false,
314
387
  "temperature": false,
315
388
  "knowledge": "2024-09-30",
316
- "release_date": "2025-10-06",
317
- "last_updated": "2025-10-06",
389
+ "release_date": "2026-07-06",
390
+ "last_updated": "2026-07-06",
318
391
  "modalities": {
319
392
  "input": [
320
393
  "text",
394
+ "audio",
321
395
  "image"
322
396
  ],
323
397
  "output": [
324
- "text"
398
+ "text",
399
+ "audio"
325
400
  ]
326
401
  },
327
402
  "open_weights": false,
328
403
  "limit": {
329
- "context": 400000,
330
- "input": 272000,
331
- "output": 272000
404
+ "context": 128000,
405
+ "input": 96000,
406
+ "output": 32000
332
407
  },
333
408
  "cost": {
334
- "input": 15,
335
- "output": 120
409
+ "input": 4,
410
+ "output": 24,
411
+ "cache_read": 0.4,
412
+ "input_audio": 32,
413
+ "output_audio": 64
336
414
  }
337
415
  },
338
- "gpt-4o": {
339
- "id": "gpt-4o",
340
- "name": "GPT-4o",
341
- "description": "Omni-era GPT for multimodal chat, practical coding, and general assistants",
416
+ "gpt-5.3-chat-latest": {
417
+ "id": "gpt-5.3-chat-latest",
418
+ "name": "GPT-5.3 Chat (latest)",
419
+ "description": "Chat-tuned GPT model for conversational assistance, writing, and tool workflows",
342
420
  "family": "gpt",
343
421
  "attachment": true,
344
422
  "reasoning": false,
345
423
  "tool_call": true,
346
424
  "structured_output": true,
347
425
  "temperature": true,
348
- "knowledge": "2023-09",
349
- "release_date": "2024-05-13",
350
- "last_updated": "2024-08-06",
426
+ "knowledge": "2025-08-31",
427
+ "release_date": "2026-03-03",
428
+ "last_updated": "2026-03-03",
351
429
  "modalities": {
352
430
  "input": [
353
431
  "text",
354
- "image",
355
- "pdf"
432
+ "image"
356
433
  ],
357
434
  "output": [
358
435
  "text"
@@ -364,27 +441,39 @@
364
441
  "output": 16384
365
442
  },
366
443
  "cost": {
367
- "input": 2.5,
368
- "output": 10,
369
- "cache_read": 1.25
444
+ "input": 1.75,
445
+ "output": 14,
446
+ "cache_read": 0.175
370
447
  }
371
448
  },
372
- "gpt-4": {
373
- "id": "gpt-4",
374
- "name": "GPT-4",
375
- "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
376
- "family": "gpt",
449
+ "o1": {
450
+ "id": "o1",
451
+ "name": "o1",
452
+ "description": "O-series reasoning model for hard analysis, math, coding, and planning",
453
+ "family": "o",
377
454
  "attachment": true,
378
- "reasoning": false,
455
+ "reasoning": true,
456
+ "reasoning_options": [
457
+ {
458
+ "type": "effort",
459
+ "values": [
460
+ "low",
461
+ "medium",
462
+ "high"
463
+ ]
464
+ }
465
+ ],
379
466
  "tool_call": true,
380
- "structured_output": false,
381
- "temperature": true,
382
- "knowledge": "2023-11",
383
- "release_date": "2023-11-06",
384
- "last_updated": "2024-04-09",
467
+ "structured_output": true,
468
+ "temperature": false,
469
+ "knowledge": "2023-09",
470
+ "release_date": "2024-12-05",
471
+ "last_updated": "2024-12-05",
385
472
  "modalities": {
386
473
  "input": [
387
- "text"
474
+ "text",
475
+ "image",
476
+ "pdf"
388
477
  ],
389
478
  "output": [
390
479
  "text"
@@ -392,25 +481,27 @@
392
481
  },
393
482
  "open_weights": false,
394
483
  "limit": {
395
- "context": 8192,
396
- "output": 8192
484
+ "context": 200000,
485
+ "output": 100000
397
486
  },
398
487
  "cost": {
399
- "input": 30,
400
- "output": 60
488
+ "input": 15,
489
+ "output": 60,
490
+ "cache_read": 7.5
401
491
  }
402
492
  },
403
- "o4-mini": {
404
- "id": "o4-mini",
405
- "name": "o4-mini",
406
- "description": "Fast o-series model for compact reasoning, coding, and tool use",
407
- "family": "o-mini",
493
+ "gpt-5-mini": {
494
+ "id": "gpt-5-mini",
495
+ "name": "GPT-5 Mini",
496
+ "description": "Small GPT-5 for responsive agents, coding help, and everyday automation",
497
+ "family": "gpt-mini",
408
498
  "attachment": true,
409
499
  "reasoning": true,
410
500
  "reasoning_options": [
411
501
  {
412
502
  "type": "effort",
413
503
  "values": [
504
+ "minimal",
414
505
  "low",
415
506
  "medium",
416
507
  "high"
@@ -420,9 +511,9 @@
420
511
  "tool_call": true,
421
512
  "structured_output": true,
422
513
  "temperature": false,
423
- "knowledge": "2024-05",
424
- "release_date": "2025-04-16",
425
- "last_updated": "2025-04-16",
514
+ "knowledge": "2024-05-30",
515
+ "release_date": "2025-08-07",
516
+ "last_updated": "2025-08-07",
426
517
  "modalities": {
427
518
  "input": [
428
519
  "text",
@@ -434,42 +525,46 @@
434
525
  },
435
526
  "open_weights": false,
436
527
  "limit": {
437
- "context": 200000,
438
- "output": 100000
528
+ "context": 400000,
529
+ "input": 272000,
530
+ "output": 128000
439
531
  },
440
532
  "cost": {
441
- "input": 1.1,
442
- "output": 4.4,
443
- "cache_read": 0.275
533
+ "input": 0.25,
534
+ "output": 2,
535
+ "cache_read": 0.025
444
536
  }
445
537
  },
446
- "o3-pro": {
447
- "id": "o3-pro",
448
- "name": "o3-pro",
449
- "description": "High-effort o3 tier for difficult technical reasoning and careful answers",
450
- "family": "o-pro",
538
+ "gpt-5.3-codex-spark": {
539
+ "id": "gpt-5.3-codex-spark",
540
+ "name": "GPT-5.3 Codex Spark",
541
+ "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
542
+ "family": "gpt-codex-spark",
451
543
  "attachment": true,
452
544
  "reasoning": true,
453
545
  "reasoning_options": [
454
546
  {
455
547
  "type": "effort",
456
548
  "values": [
549
+ "none",
457
550
  "low",
458
551
  "medium",
459
- "high"
552
+ "high",
553
+ "xhigh"
460
554
  ]
461
555
  }
462
556
  ],
463
557
  "tool_call": true,
464
558
  "structured_output": true,
465
559
  "temperature": false,
466
- "knowledge": "2024-05",
467
- "release_date": "2025-06-10",
468
- "last_updated": "2025-06-10",
560
+ "knowledge": "2025-08-31",
561
+ "release_date": "2026-02-05",
562
+ "last_updated": "2026-02-05",
469
563
  "modalities": {
470
564
  "input": [
471
565
  "text",
472
- "image"
566
+ "image",
567
+ "pdf"
473
568
  ],
474
569
  "output": [
475
570
  "text"
@@ -477,55 +572,101 @@
477
572
  },
478
573
  "open_weights": false,
479
574
  "limit": {
480
- "context": 200000,
481
- "output": 100000
575
+ "context": 128000,
576
+ "input": 100000,
577
+ "output": 32000
482
578
  },
483
579
  "cost": {
484
- "input": 20,
485
- "output": 80
580
+ "input": 1.75,
581
+ "output": 14,
582
+ "cache_read": 0.175
486
583
  }
487
584
  },
488
- "chatgpt-image-latest": {
489
- "id": "chatgpt-image-latest",
490
- "name": "chatgpt-image-latest",
491
- "description": "Image model for prompt-driven generation, editing, and visual design workflows",
492
- "family": "gpt-image",
585
+ "gpt-5.4-mini": {
586
+ "id": "gpt-5.4-mini",
587
+ "name": "GPT-5.4 mini",
588
+ "description": "Strong small GPT for coding subagents, quick tool use, and high-volume work",
589
+ "family": "gpt-mini",
493
590
  "attachment": true,
494
- "reasoning": false,
495
- "tool_call": false,
591
+ "reasoning": true,
592
+ "reasoning_options": [
593
+ {
594
+ "type": "effort",
595
+ "values": [
596
+ "none",
597
+ "low",
598
+ "medium",
599
+ "high",
600
+ "xhigh"
601
+ ]
602
+ }
603
+ ],
604
+ "tool_call": true,
605
+ "structured_output": true,
496
606
  "temperature": false,
497
- "release_date": "2025-12-16",
498
- "last_updated": "2025-12-16",
607
+ "knowledge": "2025-08-31",
608
+ "release_date": "2026-03-17",
609
+ "last_updated": "2026-03-17",
499
610
  "modalities": {
500
611
  "input": [
501
612
  "text",
502
613
  "image"
503
614
  ],
504
615
  "output": [
505
- "text",
506
- "image"
616
+ "text"
507
617
  ]
508
618
  },
509
619
  "open_weights": false,
510
620
  "limit": {
511
- "context": 0,
512
- "input": 0,
513
- "output": 0
621
+ "context": 400000,
622
+ "input": 272000,
623
+ "output": 128000
624
+ },
625
+ "experimental": {
626
+ "modes": {
627
+ "fast": {
628
+ "cost": {
629
+ "input": 1.5,
630
+ "output": 9,
631
+ "cache_read": 0.15
632
+ },
633
+ "provider": {
634
+ "body": {
635
+ "service_tier": "priority"
636
+ }
637
+ }
638
+ }
639
+ }
640
+ },
641
+ "cost": {
642
+ "input": 0.75,
643
+ "output": 4.5,
644
+ "cache_read": 0.075
514
645
  }
515
646
  },
516
- "gpt-4o-2024-05-13": {
517
- "id": "gpt-4o-2024-05-13",
518
- "name": "GPT-4o (2024-05-13)",
519
- "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
520
- "family": "gpt",
647
+ "o1-pro": {
648
+ "id": "o1-pro",
649
+ "name": "o1-pro",
650
+ "description": "O-series reasoning model for hard analysis, math, coding, and planning",
651
+ "family": "o-pro",
521
652
  "attachment": true,
522
- "reasoning": false,
653
+ "reasoning": true,
654
+ "reasoning_options": [
655
+ {
656
+ "type": "effort",
657
+ "values": [
658
+ "low",
659
+ "medium",
660
+ "high"
661
+ ]
662
+ }
663
+ ],
523
664
  "tool_call": true,
524
665
  "structured_output": true,
525
- "temperature": true,
666
+ "temperature": false,
526
667
  "knowledge": "2023-09",
527
- "release_date": "2024-05-13",
528
- "last_updated": "2024-05-13",
668
+ "release_date": "2025-03-19",
669
+ "last_updated": "2025-03-19",
529
670
  "modalities": {
530
671
  "input": [
531
672
  "text",
@@ -537,19 +678,19 @@
537
678
  },
538
679
  "open_weights": false,
539
680
  "limit": {
540
- "context": 128000,
541
- "output": 4096
681
+ "context": 200000,
682
+ "output": 100000
542
683
  },
543
684
  "cost": {
544
- "input": 5,
545
- "output": 15
685
+ "input": 150,
686
+ "output": 600
546
687
  }
547
688
  },
548
- "gpt-5.4-nano": {
549
- "id": "gpt-5.4-nano",
550
- "name": "GPT-5.4 nano",
551
- "description": "Cheapest GPT-5.4 lane for simple routing, extraction, and bulk automation",
552
- "family": "gpt-nano",
689
+ "gpt-5.6-terra": {
690
+ "id": "gpt-5.6-terra",
691
+ "name": "GPT-5.6 Terra",
692
+ "description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
693
+ "family": "gpt-terra",
553
694
  "attachment": true,
554
695
  "reasoning": true,
555
696
  "reasoning_options": [
@@ -560,20 +701,22 @@
560
701
  "low",
561
702
  "medium",
562
703
  "high",
563
- "xhigh"
704
+ "xhigh",
705
+ "max"
564
706
  ]
565
707
  }
566
708
  ],
567
709
  "tool_call": true,
568
710
  "structured_output": true,
569
711
  "temperature": false,
570
- "knowledge": "2025-08-31",
571
- "release_date": "2026-03-17",
572
- "last_updated": "2026-03-17",
712
+ "knowledge": "2026-02-16",
713
+ "release_date": "2026-07-09",
714
+ "last_updated": "2026-07-09",
573
715
  "modalities": {
574
716
  "input": [
575
717
  "text",
576
- "image"
718
+ "image",
719
+ "pdf"
577
720
  ],
578
721
  "output": [
579
722
  "text"
@@ -581,100 +724,94 @@
581
724
  },
582
725
  "open_weights": false,
583
726
  "limit": {
584
- "context": 400000,
585
- "input": 272000,
727
+ "context": 1050000,
728
+ "input": 922000,
586
729
  "output": 128000
587
730
  },
731
+ "experimental": {
732
+ "modes": {
733
+ "fast": {
734
+ "cost": {
735
+ "input": 5,
736
+ "output": 30,
737
+ "cache_read": 0.5,
738
+ "cache_write": 6.25
739
+ },
740
+ "provider": {
741
+ "body": {
742
+ "service_tier": "priority"
743
+ }
744
+ }
745
+ },
746
+ "pro": {
747
+ "provider": {
748
+ "body": {
749
+ "reasoning": {
750
+ "mode": "pro"
751
+ }
752
+ }
753
+ }
754
+ }
755
+ }
756
+ },
588
757
  "cost": {
589
- "input": 0.2,
590
- "output": 1.25,
591
- "cache_read": 0.02
758
+ "input": 2.5,
759
+ "output": 15,
760
+ "cache_read": 0.25,
761
+ "cache_write": 3.125,
762
+ "tiers": [
763
+ {
764
+ "input": 5,
765
+ "output": 22.5,
766
+ "cache_read": 0.5,
767
+ "cache_write": 6.25,
768
+ "tier": {
769
+ "type": "context",
770
+ "size": 272000
771
+ }
772
+ }
773
+ ],
774
+ "context_over_200k": {
775
+ "input": 5,
776
+ "output": 22.5,
777
+ "cache_read": 0.5,
778
+ "cache_write": 6.25
779
+ }
592
780
  }
593
781
  },
594
- "gpt-5-chat-latest": {
595
- "id": "gpt-5-chat-latest",
596
- "name": "GPT-5 Chat (latest)",
597
- "description": "Chat-tuned GPT model for conversational assistance, writing, and tool workflows",
598
- "family": "gpt-codex",
782
+ "gpt-image-1-mini": {
783
+ "id": "gpt-image-1-mini",
784
+ "name": "gpt-image-1-mini",
785
+ "description": "Image model for prompt-driven generation, editing, and visual design workflows",
786
+ "family": "gpt-image",
599
787
  "attachment": true,
600
- "reasoning": true,
601
- "reasoning_options": [],
788
+ "reasoning": false,
602
789
  "tool_call": false,
603
- "structured_output": true,
604
- "temperature": true,
605
- "knowledge": "2024-09-30",
606
- "release_date": "2025-08-07",
607
- "last_updated": "2025-08-07",
790
+ "temperature": false,
791
+ "release_date": "2025-09-26",
792
+ "last_updated": "2025-09-26",
608
793
  "modalities": {
609
794
  "input": [
610
795
  "text",
611
796
  "image"
612
797
  ],
613
798
  "output": [
614
- "text"
615
- ]
616
- },
617
- "open_weights": false,
618
- "limit": {
619
- "context": 400000,
620
- "input": 272000,
621
- "output": 128000
622
- },
623
- "cost": {
624
- "input": 1.25,
625
- "output": 10,
626
- "cache_read": 0.125
627
- }
628
- },
629
- "gpt-5.1-codex": {
630
- "id": "gpt-5.1-codex",
631
- "name": "GPT-5.1 Codex",
632
- "description": "Codex GPT for repository edits, code review, and practical software agents",
633
- "family": "gpt-codex",
634
- "attachment": true,
635
- "reasoning": true,
636
- "reasoning_options": [
637
- {
638
- "type": "effort",
639
- "values": [
640
- "low",
641
- "medium",
642
- "high"
643
- ]
644
- }
645
- ],
646
- "tool_call": true,
647
- "structured_output": true,
648
- "temperature": false,
649
- "knowledge": "2024-09-30",
650
- "release_date": "2025-11-13",
651
- "last_updated": "2025-11-13",
652
- "modalities": {
653
- "input": [
654
799
  "text",
655
800
  "image"
656
- ],
657
- "output": [
658
- "text"
659
801
  ]
660
802
  },
661
803
  "open_weights": false,
662
804
  "limit": {
663
- "context": 400000,
664
- "input": 272000,
665
- "output": 128000
666
- },
667
- "cost": {
668
- "input": 1.25,
669
- "output": 10,
670
- "cache_read": 0.125
805
+ "context": 0,
806
+ "input": 0,
807
+ "output": 0
671
808
  }
672
809
  },
673
- "gpt-5.3-codex-spark": {
674
- "id": "gpt-5.3-codex-spark",
675
- "name": "GPT-5.3 Codex Spark",
810
+ "gpt-5.3-codex": {
811
+ "id": "gpt-5.3-codex",
812
+ "name": "GPT-5.3 Codex",
676
813
  "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
677
- "family": "gpt-codex-spark",
814
+ "family": "gpt-codex",
678
815
  "attachment": true,
679
816
  "reasoning": true,
680
817
  "reasoning_options": [
@@ -707,9 +844,9 @@
707
844
  },
708
845
  "open_weights": false,
709
846
  "limit": {
710
- "context": 128000,
711
- "input": 100000,
712
- "output": 32000
847
+ "context": 400000,
848
+ "input": 272000,
849
+ "output": 128000
713
850
  },
714
851
  "cost": {
715
852
  "input": 1.75,
@@ -762,19 +899,27 @@
762
899
  "cache_read": 0.125
763
900
  }
764
901
  },
765
- "gpt-5.3-chat-latest": {
766
- "id": "gpt-5.3-chat-latest",
767
- "name": "GPT-5.3 Chat (latest)",
768
- "description": "Chat-tuned GPT model for conversational assistance, writing, and tool workflows",
769
- "family": "gpt",
902
+ "gpt-5.1-chat-latest": {
903
+ "id": "gpt-5.1-chat-latest",
904
+ "name": "GPT-5.1 Chat",
905
+ "description": "Chat-tuned GPT-5.1 for polished assistants, writing, and product conversations",
906
+ "family": "gpt-codex",
770
907
  "attachment": true,
771
- "reasoning": false,
908
+ "reasoning": true,
909
+ "reasoning_options": [
910
+ {
911
+ "type": "effort",
912
+ "values": [
913
+ "medium"
914
+ ]
915
+ }
916
+ ],
772
917
  "tool_call": true,
773
918
  "structured_output": true,
774
- "temperature": true,
775
- "knowledge": "2025-08-31",
776
- "release_date": "2026-03-03",
777
- "last_updated": "2026-03-03",
919
+ "temperature": false,
920
+ "knowledge": "2024-09-30",
921
+ "release_date": "2025-11-13",
922
+ "last_updated": "2025-11-13",
778
923
  "modalities": {
779
924
  "input": [
780
925
  "text",
@@ -790,72 +935,9 @@
790
935
  "output": 16384
791
936
  },
792
937
  "cost": {
793
- "input": 1.75,
794
- "output": 14,
795
- "cache_read": 0.175
796
- }
797
- },
798
- "gpt-4o-2024-08-06": {
799
- "id": "gpt-4o-2024-08-06",
800
- "name": "GPT-4o (2024-08-06)",
801
- "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
802
- "family": "gpt",
803
- "attachment": true,
804
- "reasoning": false,
805
- "tool_call": true,
806
- "structured_output": true,
807
- "temperature": true,
808
- "knowledge": "2023-09",
809
- "release_date": "2024-08-06",
810
- "last_updated": "2024-08-06",
811
- "modalities": {
812
- "input": [
813
- "text",
814
- "image"
815
- ],
816
- "output": [
817
- "text"
818
- ]
819
- },
820
- "open_weights": false,
821
- "limit": {
822
- "context": 128000,
823
- "output": 16384
824
- },
825
- "cost": {
826
- "input": 2.5,
938
+ "input": 1.25,
827
939
  "output": 10,
828
- "cache_read": 1.25
829
- }
830
- },
831
- "text-embedding-ada-002": {
832
- "id": "text-embedding-ada-002",
833
- "name": "text-embedding-ada-002",
834
- "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
835
- "family": "text-embedding",
836
- "attachment": false,
837
- "reasoning": false,
838
- "tool_call": false,
839
- "temperature": false,
840
- "knowledge": "2022-12",
841
- "release_date": "2022-12-15",
842
- "last_updated": "2022-12-15",
843
- "modalities": {
844
- "input": [
845
- "text"
846
- ],
847
- "output": [
848
- "text"
849
- ]
850
- },
851
- "open_weights": false,
852
- "limit": {
853
- "context": 8192,
854
- "output": 1536
855
- },
856
- "cost": {
857
- "input": 0.1,
858
- "output": 0
940
+ "cache_read": 0.125
859
941
  }
860
942
  },
861
943
  "o3-mini": {
@@ -900,31 +982,29 @@
900
982
  "cache_read": 0.55
901
983
  }
902
984
  },
903
- "gpt-5.2": {
904
- "id": "gpt-5.2",
905
- "name": "GPT-5.2",
906
- "description": "Reliable GPT generation for broad coding, writing, and tool-assisted product work",
907
- "family": "gpt",
985
+ "gpt-5.1-codex": {
986
+ "id": "gpt-5.1-codex",
987
+ "name": "GPT-5.1 Codex",
988
+ "description": "Codex GPT for repository edits, code review, and practical software agents",
989
+ "family": "gpt-codex",
908
990
  "attachment": true,
909
991
  "reasoning": true,
910
992
  "reasoning_options": [
911
993
  {
912
994
  "type": "effort",
913
995
  "values": [
914
- "none",
915
996
  "low",
916
997
  "medium",
917
- "high",
918
- "xhigh"
998
+ "high"
919
999
  ]
920
1000
  }
921
1001
  ],
922
1002
  "tool_call": true,
923
1003
  "structured_output": true,
924
1004
  "temperature": false,
925
- "knowledge": "2025-08-31",
926
- "release_date": "2025-12-11",
927
- "last_updated": "2025-12-11",
1005
+ "knowledge": "2024-09-30",
1006
+ "release_date": "2025-11-13",
1007
+ "last_updated": "2025-11-13",
928
1008
  "modalities": {
929
1009
  "input": [
930
1010
  "text",
@@ -941,16 +1021,47 @@
941
1021
  "output": 128000
942
1022
  },
943
1023
  "cost": {
944
- "input": 1.75,
945
- "output": 14,
946
- "cache_read": 0.175
1024
+ "input": 1.25,
1025
+ "output": 10,
1026
+ "cache_read": 0.125
947
1027
  }
948
1028
  },
949
- "gpt-5.3-codex": {
950
- "id": "gpt-5.3-codex",
951
- "name": "GPT-5.3 Codex",
952
- "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
953
- "family": "gpt-codex",
1029
+ "gpt-4": {
1030
+ "id": "gpt-4",
1031
+ "name": "GPT-4",
1032
+ "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
1033
+ "family": "gpt",
1034
+ "attachment": true,
1035
+ "reasoning": false,
1036
+ "tool_call": true,
1037
+ "structured_output": false,
1038
+ "temperature": true,
1039
+ "knowledge": "2023-11",
1040
+ "release_date": "2023-11-06",
1041
+ "last_updated": "2024-04-09",
1042
+ "modalities": {
1043
+ "input": [
1044
+ "text"
1045
+ ],
1046
+ "output": [
1047
+ "text"
1048
+ ]
1049
+ },
1050
+ "open_weights": false,
1051
+ "limit": {
1052
+ "context": 8192,
1053
+ "output": 8192
1054
+ },
1055
+ "cost": {
1056
+ "input": 30,
1057
+ "output": 60
1058
+ }
1059
+ },
1060
+ "gpt-5.4-nano": {
1061
+ "id": "gpt-5.4-nano",
1062
+ "name": "GPT-5.4 nano",
1063
+ "description": "Cheapest GPT-5.4 lane for simple routing, extraction, and bulk automation",
1064
+ "family": "gpt-nano",
954
1065
  "attachment": true,
955
1066
  "reasoning": true,
956
1067
  "reasoning_options": [
@@ -969,13 +1080,12 @@
969
1080
  "structured_output": true,
970
1081
  "temperature": false,
971
1082
  "knowledge": "2025-08-31",
972
- "release_date": "2026-02-05",
973
- "last_updated": "2026-02-05",
1083
+ "release_date": "2026-03-17",
1084
+ "last_updated": "2026-03-17",
974
1085
  "modalities": {
975
1086
  "input": [
976
1087
  "text",
977
- "image",
978
- "pdf"
1088
+ "image"
979
1089
  ],
980
1090
  "output": [
981
1091
  "text"
@@ -988,26 +1098,29 @@
988
1098
  "output": 128000
989
1099
  },
990
1100
  "cost": {
991
- "input": 1.75,
992
- "output": 14,
993
- "cache_read": 0.175
1101
+ "input": 0.2,
1102
+ "output": 1.25,
1103
+ "cache_read": 0.02
994
1104
  }
995
1105
  },
996
- "text-embedding-3-small": {
997
- "id": "text-embedding-3-small",
998
- "name": "text-embedding-3-small",
999
- "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
1000
- "family": "text-embedding",
1001
- "attachment": false,
1106
+ "gpt-4.1": {
1107
+ "id": "gpt-4.1",
1108
+ "name": "GPT-4.1",
1109
+ "description": "Long-lived GPT workhorse for coding, instruction following, and production apps",
1110
+ "family": "gpt",
1111
+ "attachment": true,
1002
1112
  "reasoning": false,
1003
- "tool_call": false,
1004
- "temperature": false,
1005
- "knowledge": "2024-01",
1006
- "release_date": "2024-01-25",
1007
- "last_updated": "2024-01-25",
1113
+ "tool_call": true,
1114
+ "structured_output": true,
1115
+ "temperature": true,
1116
+ "knowledge": "2024-04",
1117
+ "release_date": "2025-04-14",
1118
+ "last_updated": "2025-04-14",
1008
1119
  "modalities": {
1009
1120
  "input": [
1010
- "text"
1121
+ "text",
1122
+ "image",
1123
+ "pdf"
1011
1124
  ],
1012
1125
  "output": [
1013
1126
  "text"
@@ -1015,45 +1128,32 @@
1015
1128
  },
1016
1129
  "open_weights": false,
1017
1130
  "limit": {
1018
- "context": 8191,
1019
- "output": 1536
1131
+ "context": 1047576,
1132
+ "output": 32768
1020
1133
  },
1021
1134
  "cost": {
1022
- "input": 0.02,
1023
- "output": 0
1135
+ "input": 2,
1136
+ "output": 8,
1137
+ "cache_read": 0.5
1024
1138
  }
1025
1139
  },
1026
- "gpt-5.6-luna": {
1027
- "id": "gpt-5.6-luna",
1028
- "name": "GPT-5.6 Luna",
1029
- "description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
1030
- "family": "gpt-nano",
1140
+ "gpt-4o-2024-05-13": {
1141
+ "id": "gpt-4o-2024-05-13",
1142
+ "name": "GPT-4o (2024-05-13)",
1143
+ "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
1144
+ "family": "gpt",
1031
1145
  "attachment": true,
1032
- "reasoning": true,
1033
- "reasoning_options": [
1034
- {
1035
- "type": "effort",
1036
- "values": [
1037
- "none",
1038
- "low",
1039
- "medium",
1040
- "high",
1041
- "xhigh",
1042
- "max"
1043
- ]
1044
- }
1045
- ],
1146
+ "reasoning": false,
1046
1147
  "tool_call": true,
1047
1148
  "structured_output": true,
1048
- "temperature": false,
1049
- "knowledge": "2026-02-16",
1050
- "release_date": "2026-07-09",
1051
- "last_updated": "2026-07-09",
1149
+ "temperature": true,
1150
+ "knowledge": "2023-09",
1151
+ "release_date": "2024-05-13",
1152
+ "last_updated": "2024-05-13",
1052
1153
  "modalities": {
1053
1154
  "input": [
1054
1155
  "text",
1055
- "image",
1056
- "pdf"
1156
+ "image"
1057
1157
  ],
1058
1158
  "output": [
1059
1159
  "text"
@@ -1061,84 +1161,27 @@
1061
1161
  },
1062
1162
  "open_weights": false,
1063
1163
  "limit": {
1064
- "context": 1050000,
1065
- "input": 922000,
1066
- "output": 128000
1067
- },
1068
- "experimental": {
1069
- "modes": {
1070
- "fast": {
1071
- "cost": {
1072
- "input": 2,
1073
- "output": 12,
1074
- "cache_read": 0.2,
1075
- "cache_write": 2.5
1076
- },
1077
- "provider": {
1078
- "body": {
1079
- "service_tier": "priority"
1080
- }
1081
- }
1082
- },
1083
- "pro": {
1084
- "provider": {
1085
- "body": {
1086
- "reasoning": {
1087
- "mode": "pro"
1088
- }
1089
- }
1090
- }
1091
- }
1092
- }
1164
+ "context": 128000,
1165
+ "output": 4096
1093
1166
  },
1094
1167
  "cost": {
1095
- "input": 1,
1096
- "output": 6,
1097
- "cache_read": 0.1,
1098
- "cache_write": 1.25,
1099
- "tiers": [
1100
- {
1101
- "input": 2,
1102
- "output": 9,
1103
- "cache_read": 0.2,
1104
- "cache_write": 2.5,
1105
- "tier": {
1106
- "type": "context",
1107
- "size": 272000
1108
- }
1109
- }
1110
- ],
1111
- "context_over_200k": {
1112
- "input": 2,
1113
- "output": 9,
1114
- "cache_read": 0.2,
1115
- "cache_write": 2.5
1116
- }
1168
+ "input": 5,
1169
+ "output": 15
1117
1170
  }
1118
1171
  },
1119
- "gpt-5.1-codex-mini": {
1120
- "id": "gpt-5.1-codex-mini",
1121
- "name": "GPT-5.1 Codex mini",
1122
- "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
1123
- "family": "gpt-codex",
1172
+ "gpt-4o-2024-11-20": {
1173
+ "id": "gpt-4o-2024-11-20",
1174
+ "name": "GPT-4o (2024-11-20)",
1175
+ "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
1176
+ "family": "gpt",
1124
1177
  "attachment": true,
1125
- "reasoning": true,
1126
- "reasoning_options": [
1127
- {
1128
- "type": "effort",
1129
- "values": [
1130
- "low",
1131
- "medium",
1132
- "high"
1133
- ]
1134
- }
1135
- ],
1178
+ "reasoning": false,
1136
1179
  "tool_call": true,
1137
1180
  "structured_output": true,
1138
- "temperature": false,
1139
- "knowledge": "2024-09-30",
1140
- "release_date": "2025-11-13",
1141
- "last_updated": "2025-11-13",
1181
+ "temperature": true,
1182
+ "knowledge": "2023-09",
1183
+ "release_date": "2024-11-20",
1184
+ "last_updated": "2024-11-20",
1142
1185
  "modalities": {
1143
1186
  "input": [
1144
1187
  "text",
@@ -1150,92 +1193,132 @@
1150
1193
  },
1151
1194
  "open_weights": false,
1152
1195
  "limit": {
1153
- "context": 400000,
1154
- "input": 272000,
1155
- "output": 128000
1196
+ "context": 128000,
1197
+ "output": 16384
1156
1198
  },
1157
1199
  "cost": {
1158
- "input": 0.25,
1159
- "output": 2,
1160
- "cache_read": 0.025
1200
+ "input": 2.5,
1201
+ "output": 10,
1202
+ "cache_read": 1.25
1161
1203
  }
1162
1204
  },
1163
- "gpt-realtime-2.1": {
1164
- "id": "gpt-realtime-2.1",
1165
- "name": "GPT-Realtime-2.1",
1166
- "description": "Realtime speech-to-speech model with configurable reasoning, tool use, and robust voice-agent behavior",
1205
+ "gpt-3.5-turbo": {
1206
+ "id": "gpt-3.5-turbo",
1207
+ "name": "GPT-3.5-turbo",
1208
+ "description": "Compact GPT model for low-latency assistance and high-volume workloads",
1167
1209
  "family": "gpt",
1210
+ "attachment": false,
1211
+ "reasoning": false,
1212
+ "tool_call": false,
1213
+ "structured_output": false,
1214
+ "temperature": true,
1215
+ "knowledge": "2021-09-01",
1216
+ "release_date": "2023-03-01",
1217
+ "last_updated": "2023-11-06",
1218
+ "modalities": {
1219
+ "input": [
1220
+ "text"
1221
+ ],
1222
+ "output": [
1223
+ "text"
1224
+ ]
1225
+ },
1226
+ "open_weights": false,
1227
+ "limit": {
1228
+ "context": 16385,
1229
+ "output": 4096
1230
+ },
1231
+ "cost": {
1232
+ "input": 0.5,
1233
+ "output": 1.5,
1234
+ "cache_read": 0
1235
+ }
1236
+ },
1237
+ "gpt-4.1-nano": {
1238
+ "id": "gpt-4.1-nano",
1239
+ "name": "GPT-4.1 nano",
1240
+ "description": "Tiny GPT-4.1 option for classification, routing, and very high-volume tasks",
1241
+ "family": "gpt-nano",
1168
1242
  "attachment": true,
1169
- "reasoning": true,
1170
- "reasoning_options": [
1171
- {
1172
- "type": "effort",
1173
- "values": [
1174
- "minimal",
1175
- "low",
1176
- "medium",
1177
- "high",
1178
- "xhigh"
1179
- ]
1180
- }
1181
- ],
1243
+ "reasoning": false,
1182
1244
  "tool_call": true,
1183
- "structured_output": false,
1184
- "temperature": false,
1185
- "knowledge": "2024-09-30",
1186
- "release_date": "2026-07-06",
1187
- "last_updated": "2026-07-06",
1245
+ "structured_output": true,
1246
+ "temperature": true,
1247
+ "knowledge": "2024-04",
1248
+ "release_date": "2025-04-14",
1249
+ "last_updated": "2025-04-14",
1188
1250
  "modalities": {
1189
1251
  "input": [
1190
1252
  "text",
1191
- "audio",
1192
1253
  "image"
1193
1254
  ],
1194
1255
  "output": [
1195
- "text",
1196
- "audio"
1256
+ "text"
1197
1257
  ]
1198
1258
  },
1199
1259
  "open_weights": false,
1200
1260
  "limit": {
1201
- "context": 128000,
1202
- "input": 96000,
1203
- "output": 32000
1261
+ "context": 1047576,
1262
+ "output": 32768
1204
1263
  },
1205
1264
  "cost": {
1206
- "input": 4,
1207
- "output": 24,
1208
- "cache_read": 0.4,
1209
- "input_audio": 32,
1210
- "output_audio": 64
1265
+ "input": 0.1,
1266
+ "output": 0.4,
1267
+ "cache_read": 0.025
1211
1268
  }
1212
1269
  },
1213
- "gpt-5.6-terra": {
1214
- "id": "gpt-5.6-terra",
1215
- "name": "GPT-5.6 Terra",
1216
- "description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
1217
- "family": "gpt-mini",
1270
+ "gpt-image-1.5": {
1271
+ "id": "gpt-image-1.5",
1272
+ "name": "gpt-image-1.5",
1273
+ "description": "Image model for prompt-driven generation, editing, and visual design workflows",
1274
+ "family": "gpt-image",
1275
+ "attachment": true,
1276
+ "reasoning": false,
1277
+ "tool_call": false,
1278
+ "temperature": false,
1279
+ "release_date": "2025-11-25",
1280
+ "last_updated": "2025-11-25",
1281
+ "modalities": {
1282
+ "input": [
1283
+ "text",
1284
+ "image"
1285
+ ],
1286
+ "output": [
1287
+ "text",
1288
+ "image"
1289
+ ]
1290
+ },
1291
+ "open_weights": false,
1292
+ "limit": {
1293
+ "context": 0,
1294
+ "input": 0,
1295
+ "output": 0
1296
+ }
1297
+ },
1298
+ "gpt-5.2-codex": {
1299
+ "id": "gpt-5.2-codex",
1300
+ "name": "GPT-5.2 Codex",
1301
+ "description": "Code-specialist GPT for repository edits, reviews, and long-running software agents",
1302
+ "family": "gpt-codex",
1218
1303
  "attachment": true,
1219
1304
  "reasoning": true,
1220
1305
  "reasoning_options": [
1221
1306
  {
1222
1307
  "type": "effort",
1223
1308
  "values": [
1224
- "none",
1225
1309
  "low",
1226
1310
  "medium",
1227
1311
  "high",
1228
- "xhigh",
1229
- "max"
1312
+ "xhigh"
1230
1313
  ]
1231
1314
  }
1232
1315
  ],
1233
1316
  "tool_call": true,
1234
1317
  "structured_output": true,
1235
1318
  "temperature": false,
1236
- "knowledge": "2026-02-16",
1237
- "release_date": "2026-07-09",
1238
- "last_updated": "2026-07-09",
1319
+ "knowledge": "2025-08-31",
1320
+ "release_date": "2025-12-11",
1321
+ "last_updated": "2025-12-11",
1239
1322
  "modalities": {
1240
1323
  "input": [
1241
1324
  "text",
@@ -1248,65 +1331,50 @@
1248
1331
  },
1249
1332
  "open_weights": false,
1250
1333
  "limit": {
1251
- "context": 1050000,
1252
- "input": 922000,
1334
+ "context": 400000,
1335
+ "input": 272000,
1253
1336
  "output": 128000
1254
1337
  },
1255
- "experimental": {
1256
- "modes": {
1257
- "fast": {
1258
- "cost": {
1259
- "input": 5,
1260
- "output": 30,
1261
- "cache_read": 0.5,
1262
- "cache_write": 6.25
1263
- },
1264
- "provider": {
1265
- "body": {
1266
- "service_tier": "priority"
1267
- }
1268
- }
1269
- },
1270
- "pro": {
1271
- "provider": {
1272
- "body": {
1273
- "reasoning": {
1274
- "mode": "pro"
1275
- }
1276
- }
1277
- }
1278
- }
1279
- }
1280
- },
1281
1338
  "cost": {
1282
- "input": 2.5,
1283
- "output": 15,
1284
- "cache_read": 0.25,
1285
- "cache_write": 3.125,
1286
- "tiers": [
1287
- {
1288
- "input": 5,
1289
- "output": 22.5,
1290
- "cache_read": 0.5,
1291
- "cache_write": 6.25,
1292
- "tier": {
1293
- "type": "context",
1294
- "size": 272000
1295
- }
1296
- }
1339
+ "input": 1.75,
1340
+ "output": 14,
1341
+ "cache_read": 0.175
1342
+ }
1343
+ },
1344
+ "text-embedding-3-large": {
1345
+ "id": "text-embedding-3-large",
1346
+ "name": "text-embedding-3-large",
1347
+ "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
1348
+ "family": "text-embedding",
1349
+ "attachment": false,
1350
+ "reasoning": false,
1351
+ "tool_call": false,
1352
+ "temperature": false,
1353
+ "knowledge": "2024-01",
1354
+ "release_date": "2024-01-25",
1355
+ "last_updated": "2024-01-25",
1356
+ "modalities": {
1357
+ "input": [
1358
+ "text"
1297
1359
  ],
1298
- "context_over_200k": {
1299
- "input": 5,
1300
- "output": 22.5,
1301
- "cache_read": 0.5,
1302
- "cache_write": 6.25
1303
- }
1360
+ "output": [
1361
+ "text"
1362
+ ]
1363
+ },
1364
+ "open_weights": false,
1365
+ "limit": {
1366
+ "context": 8191,
1367
+ "output": 3072
1368
+ },
1369
+ "cost": {
1370
+ "input": 0.13,
1371
+ "output": 0
1304
1372
  }
1305
1373
  },
1306
- "gpt-5.1-chat-latest": {
1307
- "id": "gpt-5.1-chat-latest",
1308
- "name": "GPT-5.1 Chat",
1309
- "description": "Chat-tuned GPT-5.1 for polished assistants, writing, and product conversations",
1374
+ "gpt-5.1-codex-mini": {
1375
+ "id": "gpt-5.1-codex-mini",
1376
+ "name": "GPT-5.1 Codex mini",
1377
+ "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
1310
1378
  "family": "gpt-codex",
1311
1379
  "attachment": true,
1312
1380
  "reasoning": true,
@@ -1314,7 +1382,9 @@
1314
1382
  {
1315
1383
  "type": "effort",
1316
1384
  "values": [
1317
- "medium"
1385
+ "low",
1386
+ "medium",
1387
+ "high"
1318
1388
  ]
1319
1389
  }
1320
1390
  ],
@@ -1335,13 +1405,14 @@
1335
1405
  },
1336
1406
  "open_weights": false,
1337
1407
  "limit": {
1338
- "context": 128000,
1339
- "output": 16384
1408
+ "context": 400000,
1409
+ "input": 272000,
1410
+ "output": 128000
1340
1411
  },
1341
1412
  "cost": {
1342
- "input": 1.25,
1343
- "output": 10,
1344
- "cache_read": 0.125
1413
+ "input": 0.25,
1414
+ "output": 2,
1415
+ "cache_read": 0.025
1345
1416
  }
1346
1417
  },
1347
1418
  "gpt-5.2-chat-latest": {
@@ -1385,26 +1456,30 @@
1385
1456
  "cache_read": 0.175
1386
1457
  }
1387
1458
  },
1388
- "o4-mini-deep-research": {
1389
- "id": "o4-mini-deep-research",
1390
- "name": "o4-mini-deep-research",
1391
- "description": "Research model for long-horizon investigation, synthesis, and analytical reports",
1392
- "family": "o-mini",
1459
+ "gpt-5": {
1460
+ "id": "gpt-5",
1461
+ "name": "GPT-5",
1462
+ "description": "Original GPT-5 workhorse for reasoning, coding, writing, and tool workflows",
1463
+ "family": "gpt",
1393
1464
  "attachment": true,
1394
1465
  "reasoning": true,
1395
1466
  "reasoning_options": [
1396
1467
  {
1397
1468
  "type": "effort",
1398
1469
  "values": [
1399
- "medium"
1470
+ "minimal",
1471
+ "low",
1472
+ "medium",
1473
+ "high"
1400
1474
  ]
1401
1475
  }
1402
1476
  ],
1403
1477
  "tool_call": true,
1478
+ "structured_output": true,
1404
1479
  "temperature": false,
1405
- "knowledge": "2024-05",
1406
- "release_date": "2024-06-26",
1407
- "last_updated": "2024-06-26",
1480
+ "knowledge": "2024-09-30",
1481
+ "release_date": "2025-08-07",
1482
+ "last_updated": "2025-08-07",
1408
1483
  "modalities": {
1409
1484
  "input": [
1410
1485
  "text",
@@ -1416,60 +1491,66 @@
1416
1491
  },
1417
1492
  "open_weights": false,
1418
1493
  "limit": {
1419
- "context": 200000,
1420
- "output": 100000
1494
+ "context": 400000,
1495
+ "input": 272000,
1496
+ "output": 128000
1421
1497
  },
1422
1498
  "cost": {
1423
- "input": 2,
1424
- "output": 8,
1425
- "cache_read": 0.5
1499
+ "input": 1.25,
1500
+ "output": 10,
1501
+ "cache_read": 0.125
1426
1502
  }
1427
1503
  },
1428
- "gpt-image-1.5": {
1429
- "id": "gpt-image-1.5",
1430
- "name": "gpt-image-1.5",
1431
- "description": "Image model for prompt-driven generation, editing, and visual design workflows",
1432
- "family": "gpt-image",
1504
+ "gpt-5-chat-latest": {
1505
+ "id": "gpt-5-chat-latest",
1506
+ "name": "GPT-5 Chat (latest)",
1507
+ "description": "Chat-tuned GPT model for conversational assistance, writing, and tool workflows",
1508
+ "family": "gpt-codex",
1433
1509
  "attachment": true,
1434
- "reasoning": false,
1510
+ "reasoning": true,
1511
+ "reasoning_options": [],
1435
1512
  "tool_call": false,
1436
- "temperature": false,
1437
- "release_date": "2025-11-25",
1438
- "last_updated": "2025-11-25",
1513
+ "structured_output": true,
1514
+ "temperature": true,
1515
+ "knowledge": "2024-09-30",
1516
+ "release_date": "2025-08-07",
1517
+ "last_updated": "2025-08-07",
1439
1518
  "modalities": {
1440
1519
  "input": [
1441
1520
  "text",
1442
1521
  "image"
1443
1522
  ],
1444
1523
  "output": [
1445
- "text",
1446
- "image"
1524
+ "text"
1447
1525
  ]
1448
1526
  },
1449
1527
  "open_weights": false,
1450
1528
  "limit": {
1451
- "context": 0,
1452
- "input": 0,
1453
- "output": 0
1529
+ "context": 400000,
1530
+ "input": 272000,
1531
+ "output": 128000
1532
+ },
1533
+ "cost": {
1534
+ "input": 1.25,
1535
+ "output": 10,
1536
+ "cache_read": 0.125
1454
1537
  }
1455
1538
  },
1456
- "gpt-4.1-nano": {
1457
- "id": "gpt-4.1-nano",
1458
- "name": "GPT-4.1 nano",
1459
- "description": "Tiny GPT-4.1 option for classification, routing, and very high-volume tasks",
1460
- "family": "gpt-nano",
1461
- "attachment": true,
1539
+ "text-embedding-ada-002": {
1540
+ "id": "text-embedding-ada-002",
1541
+ "name": "text-embedding-ada-002",
1542
+ "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
1543
+ "family": "text-embedding",
1544
+ "attachment": false,
1462
1545
  "reasoning": false,
1463
- "tool_call": true,
1464
- "structured_output": true,
1465
- "temperature": true,
1466
- "knowledge": "2024-04",
1467
- "release_date": "2025-04-14",
1468
- "last_updated": "2025-04-14",
1546
+ "tool_call": false,
1547
+ "temperature": false,
1548
+ "knowledge": "2022-12",
1549
+ "release_date": "2022-12-15",
1550
+ "last_updated": "2022-12-15",
1469
1551
  "modalities": {
1470
1552
  "input": [
1471
- "text",
1472
- "image"
1553
+ "text"
1473
1554
  ],
1474
1555
  "output": [
1475
1556
  "text"
@@ -1477,32 +1558,29 @@
1477
1558
  },
1478
1559
  "open_weights": false,
1479
1560
  "limit": {
1480
- "context": 1047576,
1481
- "output": 32768
1561
+ "context": 8192,
1562
+ "output": 1536
1482
1563
  },
1483
1564
  "cost": {
1484
1565
  "input": 0.1,
1485
- "output": 0.4,
1486
- "cache_read": 0.025
1566
+ "output": 0
1487
1567
  }
1488
1568
  },
1489
- "gpt-4o-2024-11-20": {
1490
- "id": "gpt-4o-2024-11-20",
1491
- "name": "GPT-4o (2024-11-20)",
1492
- "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
1493
- "family": "gpt",
1494
- "attachment": true,
1569
+ "text-embedding-3-small": {
1570
+ "id": "text-embedding-3-small",
1571
+ "name": "text-embedding-3-small",
1572
+ "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
1573
+ "family": "text-embedding",
1574
+ "attachment": false,
1495
1575
  "reasoning": false,
1496
- "tool_call": true,
1497
- "structured_output": true,
1498
- "temperature": true,
1499
- "knowledge": "2023-09",
1500
- "release_date": "2024-11-20",
1501
- "last_updated": "2024-11-20",
1576
+ "tool_call": false,
1577
+ "temperature": false,
1578
+ "knowledge": "2024-01",
1579
+ "release_date": "2024-01-25",
1580
+ "last_updated": "2024-01-25",
1502
1581
  "modalities": {
1503
1582
  "input": [
1504
- "text",
1505
- "image"
1583
+ "text"
1506
1584
  ],
1507
1585
  "output": [
1508
1586
  "text"
@@ -1510,38 +1588,27 @@
1510
1588
  },
1511
1589
  "open_weights": false,
1512
1590
  "limit": {
1513
- "context": 128000,
1514
- "output": 16384
1591
+ "context": 8191,
1592
+ "output": 1536
1515
1593
  },
1516
1594
  "cost": {
1517
- "input": 2.5,
1518
- "output": 10,
1519
- "cache_read": 1.25
1595
+ "input": 0.02,
1596
+ "output": 0
1520
1597
  }
1521
1598
  },
1522
- "o1": {
1523
- "id": "o1",
1524
- "name": "o1",
1525
- "description": "O-series reasoning model for hard analysis, math, coding, and planning",
1526
- "family": "o",
1599
+ "gpt-4o-mini": {
1600
+ "id": "gpt-4o-mini",
1601
+ "name": "GPT-4o mini",
1602
+ "description": "Small omni GPT for cheap multimodal assistance and production-scale traffic",
1603
+ "family": "gpt-mini",
1527
1604
  "attachment": true,
1528
- "reasoning": true,
1529
- "reasoning_options": [
1530
- {
1531
- "type": "effort",
1532
- "values": [
1533
- "low",
1534
- "medium",
1535
- "high"
1536
- ]
1537
- }
1538
- ],
1605
+ "reasoning": false,
1539
1606
  "tool_call": true,
1540
1607
  "structured_output": true,
1541
- "temperature": false,
1608
+ "temperature": true,
1542
1609
  "knowledge": "2023-09",
1543
- "release_date": "2024-12-05",
1544
- "last_updated": "2024-12-05",
1610
+ "release_date": "2024-07-18",
1611
+ "last_updated": "2024-07-18",
1545
1612
  "modalities": {
1546
1613
  "input": [
1547
1614
  "text",
@@ -1554,26 +1621,27 @@
1554
1621
  },
1555
1622
  "open_weights": false,
1556
1623
  "limit": {
1557
- "context": 200000,
1558
- "output": 100000
1624
+ "context": 128000,
1625
+ "output": 16384
1559
1626
  },
1560
1627
  "cost": {
1561
- "input": 15,
1562
- "output": 60,
1563
- "cache_read": 7.5
1628
+ "input": 0.15,
1629
+ "output": 0.6,
1630
+ "cache_read": 0.075
1564
1631
  }
1565
1632
  },
1566
- "o1-pro": {
1567
- "id": "o1-pro",
1568
- "name": "o1-pro",
1569
- "description": "O-series reasoning model for hard analysis, math, coding, and planning",
1570
- "family": "o-pro",
1633
+ "gpt-5.1": {
1634
+ "id": "gpt-5.1",
1635
+ "name": "GPT-5.1",
1636
+ "description": "Sharper GPT-5 generation for coding, product work, and tool-assisted tasks",
1637
+ "family": "gpt",
1571
1638
  "attachment": true,
1572
1639
  "reasoning": true,
1573
1640
  "reasoning_options": [
1574
1641
  {
1575
1642
  "type": "effort",
1576
1643
  "values": [
1644
+ "none",
1577
1645
  "low",
1578
1646
  "medium",
1579
1647
  "high"
@@ -1583,9 +1651,9 @@
1583
1651
  "tool_call": true,
1584
1652
  "structured_output": true,
1585
1653
  "temperature": false,
1586
- "knowledge": "2023-09",
1587
- "release_date": "2025-03-19",
1588
- "last_updated": "2025-03-19",
1654
+ "knowledge": "2024-09-30",
1655
+ "release_date": "2025-11-13",
1656
+ "last_updated": "2025-11-13",
1589
1657
  "modalities": {
1590
1658
  "input": [
1591
1659
  "text",
@@ -1597,19 +1665,21 @@
1597
1665
  },
1598
1666
  "open_weights": false,
1599
1667
  "limit": {
1600
- "context": 200000,
1601
- "output": 100000
1668
+ "context": 400000,
1669
+ "input": 272000,
1670
+ "output": 128000
1602
1671
  },
1603
1672
  "cost": {
1604
- "input": 150,
1605
- "output": 600
1673
+ "input": 1.25,
1674
+ "output": 10,
1675
+ "cache_read": 0.125
1606
1676
  }
1607
1677
  },
1608
- "gpt-5.4": {
1609
- "id": "gpt-5.4",
1610
- "name": "GPT-5.4",
1611
- "description": "Agent-ready GPT for coding and computer-use workflows at a lower cost",
1612
- "family": "gpt",
1678
+ "gpt-5.6": {
1679
+ "id": "gpt-5.6",
1680
+ "name": "GPT-5.6",
1681
+ "description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
1682
+ "family": "gpt-sol",
1613
1683
  "attachment": true,
1614
1684
  "reasoning": true,
1615
1685
  "reasoning_options": [
@@ -1620,16 +1690,17 @@
1620
1690
  "low",
1621
1691
  "medium",
1622
1692
  "high",
1623
- "xhigh"
1693
+ "xhigh",
1694
+ "max"
1624
1695
  ]
1625
1696
  }
1626
1697
  ],
1627
1698
  "tool_call": true,
1628
1699
  "structured_output": true,
1629
1700
  "temperature": false,
1630
- "knowledge": "2025-08-31",
1631
- "release_date": "2026-03-05",
1632
- "last_updated": "2026-03-05",
1701
+ "knowledge": "2026-02-16",
1702
+ "release_date": "2026-07-09",
1703
+ "last_updated": "2026-07-09",
1633
1704
  "modalities": {
1634
1705
  "input": [
1635
1706
  "text",
@@ -1650,27 +1721,39 @@
1650
1721
  "modes": {
1651
1722
  "fast": {
1652
1723
  "cost": {
1653
- "input": 5,
1654
- "output": 30,
1655
- "cache_read": 0.5
1724
+ "input": 10,
1725
+ "output": 60,
1726
+ "cache_read": 1,
1727
+ "cache_write": 12.5
1656
1728
  },
1657
1729
  "provider": {
1658
1730
  "body": {
1659
1731
  "service_tier": "priority"
1660
1732
  }
1661
1733
  }
1734
+ },
1735
+ "pro": {
1736
+ "provider": {
1737
+ "body": {
1738
+ "reasoning": {
1739
+ "mode": "pro"
1740
+ }
1741
+ }
1742
+ }
1662
1743
  }
1663
1744
  }
1664
1745
  },
1665
1746
  "cost": {
1666
- "input": 2.5,
1667
- "output": 15,
1668
- "cache_read": 0.25,
1747
+ "input": 5,
1748
+ "output": 30,
1749
+ "cache_read": 0.5,
1750
+ "cache_write": 6.25,
1669
1751
  "tiers": [
1670
1752
  {
1671
- "input": 5,
1672
- "output": 22.5,
1673
- "cache_read": 0.5,
1753
+ "input": 10,
1754
+ "output": 45,
1755
+ "cache_read": 1,
1756
+ "cache_write": 12.5,
1674
1757
  "tier": {
1675
1758
  "type": "context",
1676
1759
  "size": 272000
@@ -1678,37 +1761,33 @@
1678
1761
  }
1679
1762
  ],
1680
1763
  "context_over_200k": {
1681
- "input": 5,
1682
- "output": 22.5,
1683
- "cache_read": 0.5
1764
+ "input": 10,
1765
+ "output": 45,
1766
+ "cache_read": 1,
1767
+ "cache_write": 12.5
1684
1768
  }
1685
1769
  }
1686
1770
  },
1687
- "gpt-5.4-mini": {
1688
- "id": "gpt-5.4-mini",
1689
- "name": "GPT-5.4 mini",
1690
- "description": "Strong small GPT for coding subagents, quick tool use, and high-volume work",
1691
- "family": "gpt-mini",
1771
+ "o3-deep-research": {
1772
+ "id": "o3-deep-research",
1773
+ "name": "o3-deep-research",
1774
+ "description": "Research model for long-horizon investigation, synthesis, and analytical reports",
1775
+ "family": "o",
1692
1776
  "attachment": true,
1693
1777
  "reasoning": true,
1694
1778
  "reasoning_options": [
1695
1779
  {
1696
1780
  "type": "effort",
1697
1781
  "values": [
1698
- "none",
1699
- "low",
1700
- "medium",
1701
- "high",
1702
- "xhigh"
1782
+ "medium"
1703
1783
  ]
1704
1784
  }
1705
1785
  ],
1706
1786
  "tool_call": true,
1707
- "structured_output": true,
1708
1787
  "temperature": false,
1709
- "knowledge": "2025-08-31",
1710
- "release_date": "2026-03-17",
1711
- "last_updated": "2026-03-17",
1788
+ "knowledge": "2024-05",
1789
+ "release_date": "2024-06-26",
1790
+ "last_updated": "2024-06-26",
1712
1791
  "modalities": {
1713
1792
  "input": [
1714
1793
  "text",
@@ -1720,86 +1799,28 @@
1720
1799
  },
1721
1800
  "open_weights": false,
1722
1801
  "limit": {
1723
- "context": 400000,
1724
- "input": 272000,
1725
- "output": 128000
1726
- },
1727
- "experimental": {
1728
- "modes": {
1729
- "fast": {
1730
- "cost": {
1731
- "input": 1.5,
1732
- "output": 9,
1733
- "cache_read": 0.15
1734
- },
1735
- "provider": {
1736
- "body": {
1737
- "service_tier": "priority"
1738
- }
1739
- }
1740
- }
1741
- }
1802
+ "context": 200000,
1803
+ "output": 100000
1742
1804
  },
1743
1805
  "cost": {
1744
- "input": 0.75,
1745
- "output": 4.5,
1746
- "cache_read": 0.075
1806
+ "input": 10,
1807
+ "output": 40,
1808
+ "cache_read": 2.5
1747
1809
  }
1748
1810
  },
1749
- "gpt-4.1": {
1750
- "id": "gpt-4.1",
1751
- "name": "GPT-4.1",
1752
- "description": "Long-lived GPT workhorse for coding, instruction following, and production apps",
1811
+ "gpt-4o-2024-08-06": {
1812
+ "id": "gpt-4o-2024-08-06",
1813
+ "name": "GPT-4o (2024-08-06)",
1814
+ "description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
1753
1815
  "family": "gpt",
1754
1816
  "attachment": true,
1755
1817
  "reasoning": false,
1756
1818
  "tool_call": true,
1757
1819
  "structured_output": true,
1758
1820
  "temperature": true,
1759
- "knowledge": "2024-04",
1760
- "release_date": "2025-04-14",
1761
- "last_updated": "2025-04-14",
1762
- "modalities": {
1763
- "input": [
1764
- "text",
1765
- "image",
1766
- "pdf"
1767
- ],
1768
- "output": [
1769
- "text"
1770
- ]
1771
- },
1772
- "open_weights": false,
1773
- "limit": {
1774
- "context": 1047576,
1775
- "output": 32768
1776
- },
1777
- "cost": {
1778
- "input": 2,
1779
- "output": 8,
1780
- "cache_read": 0.5
1781
- }
1782
- },
1783
- "o3-deep-research": {
1784
- "id": "o3-deep-research",
1785
- "name": "o3-deep-research",
1786
- "description": "Research model for long-horizon investigation, synthesis, and analytical reports",
1787
- "family": "o",
1788
- "attachment": true,
1789
- "reasoning": true,
1790
- "reasoning_options": [
1791
- {
1792
- "type": "effort",
1793
- "values": [
1794
- "medium"
1795
- ]
1796
- }
1797
- ],
1798
- "tool_call": true,
1799
- "temperature": false,
1800
- "knowledge": "2024-05",
1801
- "release_date": "2024-06-26",
1802
- "last_updated": "2024-06-26",
1821
+ "knowledge": "2023-09",
1822
+ "release_date": "2024-08-06",
1823
+ "last_updated": "2024-08-06",
1803
1824
  "modalities": {
1804
1825
  "input": [
1805
1826
  "text",
@@ -1811,20 +1832,20 @@
1811
1832
  },
1812
1833
  "open_weights": false,
1813
1834
  "limit": {
1814
- "context": 200000,
1815
- "output": 100000
1835
+ "context": 128000,
1836
+ "output": 16384
1816
1837
  },
1817
1838
  "cost": {
1818
- "input": 10,
1819
- "output": 40,
1820
- "cache_read": 2.5
1839
+ "input": 2.5,
1840
+ "output": 10,
1841
+ "cache_read": 1.25
1821
1842
  }
1822
1843
  },
1823
- "gpt-5-mini": {
1824
- "id": "gpt-5-mini",
1825
- "name": "GPT-5 Mini",
1826
- "description": "Small GPT-5 for responsive agents, coding help, and everyday automation",
1827
- "family": "gpt-mini",
1844
+ "gpt-5-nano": {
1845
+ "id": "gpt-5-nano",
1846
+ "name": "GPT-5 Nano",
1847
+ "description": "Tiny GPT-5 lane for routing, extraction, classification, and bulk jobs",
1848
+ "family": "gpt-nano",
1828
1849
  "attachment": true,
1829
1850
  "reasoning": true,
1830
1851
  "reasoning_options": [
@@ -1860,56 +1881,38 @@
1860
1881
  "output": 128000
1861
1882
  },
1862
1883
  "cost": {
1863
- "input": 0.25,
1864
- "output": 2,
1865
- "cache_read": 0.025
1866
- }
1867
- },
1868
- "gpt-image-1": {
1869
- "id": "gpt-image-1",
1870
- "name": "gpt-image-1",
1871
- "description": "OpenAI image model for production generation, edits, and brand-safe visual workflows",
1872
- "family": "gpt-image",
1873
- "attachment": true,
1874
- "reasoning": false,
1875
- "tool_call": false,
1876
- "temperature": false,
1877
- "release_date": "2025-04-24",
1878
- "last_updated": "2025-04-24",
1879
- "modalities": {
1880
- "input": [
1881
- "text",
1882
- "image"
1883
- ],
1884
- "output": [
1885
- "image"
1886
- ]
1887
- },
1888
- "open_weights": false,
1889
- "limit": {
1890
- "context": 0,
1891
- "input": 0,
1892
- "output": 0
1884
+ "input": 0.05,
1885
+ "output": 0.4,
1886
+ "cache_read": 0.005
1893
1887
  }
1894
1888
  },
1895
- "gpt-4.1-mini": {
1896
- "id": "gpt-4.1-mini",
1897
- "name": "GPT-4.1 mini",
1898
- "description": "Affordable GPT-4.1 lane for fast coding help and structured extraction",
1899
- "family": "gpt-mini",
1889
+ "o4-mini": {
1890
+ "id": "o4-mini",
1891
+ "name": "o4-mini",
1892
+ "description": "Fast o-series model for compact reasoning, coding, and tool use",
1893
+ "family": "o-mini",
1900
1894
  "attachment": true,
1901
- "reasoning": false,
1895
+ "reasoning": true,
1896
+ "reasoning_options": [
1897
+ {
1898
+ "type": "effort",
1899
+ "values": [
1900
+ "low",
1901
+ "medium",
1902
+ "high"
1903
+ ]
1904
+ }
1905
+ ],
1902
1906
  "tool_call": true,
1903
1907
  "structured_output": true,
1904
- "temperature": true,
1905
- "knowledge": "2024-04",
1906
- "release_date": "2025-04-14",
1907
- "last_updated": "2025-04-14",
1908
+ "temperature": false,
1909
+ "knowledge": "2024-05",
1910
+ "release_date": "2025-04-16",
1911
+ "last_updated": "2025-04-16",
1908
1912
  "modalities": {
1909
1913
  "input": [
1910
1914
  "text",
1911
- "image",
1912
- "pdf"
1915
+ "image"
1913
1916
  ],
1914
1917
  "output": [
1915
1918
  "text"
@@ -1917,13 +1920,13 @@
1917
1920
  },
1918
1921
  "open_weights": false,
1919
1922
  "limit": {
1920
- "context": 1047576,
1921
- "output": 32768
1923
+ "context": 200000,
1924
+ "output": 100000
1922
1925
  },
1923
1926
  "cost": {
1924
- "input": 0.4,
1925
- "output": 1.6,
1926
- "cache_read": 0.1
1927
+ "input": 1.1,
1928
+ "output": 4.4,
1929
+ "cache_read": 0.275
1927
1930
  }
1928
1931
  },
1929
1932
  "gpt-4-turbo": {
@@ -1958,58 +1961,31 @@
1958
1961
  "output": 30
1959
1962
  }
1960
1963
  },
1961
- "gpt-image-1-mini": {
1962
- "id": "gpt-image-1-mini",
1963
- "name": "gpt-image-1-mini",
1964
- "description": "Image model for prompt-driven generation, editing, and visual design workflows",
1965
- "family": "gpt-image",
1966
- "attachment": true,
1967
- "reasoning": false,
1968
- "tool_call": false,
1969
- "temperature": false,
1970
- "release_date": "2025-09-26",
1971
- "last_updated": "2025-09-26",
1972
- "modalities": {
1973
- "input": [
1974
- "text",
1975
- "image"
1976
- ],
1977
- "output": [
1978
- "text",
1979
- "image"
1980
- ]
1981
- },
1982
- "open_weights": false,
1983
- "limit": {
1984
- "context": 0,
1985
- "input": 0,
1986
- "output": 0
1987
- }
1988
- },
1989
- "gpt-5-nano": {
1990
- "id": "gpt-5-nano",
1991
- "name": "GPT-5 Nano",
1992
- "description": "Tiny GPT-5 lane for routing, extraction, classification, and bulk jobs",
1993
- "family": "gpt-nano",
1964
+ "gpt-5.2": {
1965
+ "id": "gpt-5.2",
1966
+ "name": "GPT-5.2",
1967
+ "description": "Reliable GPT generation for broad coding, writing, and tool-assisted product work",
1968
+ "family": "gpt",
1994
1969
  "attachment": true,
1995
1970
  "reasoning": true,
1996
1971
  "reasoning_options": [
1997
1972
  {
1998
1973
  "type": "effort",
1999
1974
  "values": [
2000
- "minimal",
1975
+ "none",
2001
1976
  "low",
2002
1977
  "medium",
2003
- "high"
1978
+ "high",
1979
+ "xhigh"
2004
1980
  ]
2005
1981
  }
2006
1982
  ],
2007
1983
  "tool_call": true,
2008
1984
  "structured_output": true,
2009
1985
  "temperature": false,
2010
- "knowledge": "2024-05-30",
2011
- "release_date": "2025-08-07",
2012
- "last_updated": "2025-08-07",
1986
+ "knowledge": "2025-08-31",
1987
+ "release_date": "2025-12-11",
1988
+ "last_updated": "2025-12-11",
2013
1989
  "modalities": {
2014
1990
  "input": [
2015
1991
  "text",
@@ -2026,38 +2002,76 @@
2026
2002
  "output": 128000
2027
2003
  },
2028
2004
  "cost": {
2029
- "input": 0.05,
2030
- "output": 0.4,
2031
- "cache_read": 0.005
2005
+ "input": 1.75,
2006
+ "output": 14,
2007
+ "cache_read": 0.175
2032
2008
  }
2033
2009
  },
2034
- "gpt-5.4-pro": {
2035
- "id": "gpt-5.4-pro",
2036
- "name": "GPT-5.4 Pro",
2037
- "description": "More exact GPT-5.4 tier for demanding professional reasoning and agent tasks",
2038
- "family": "gpt-pro",
2010
+ "gpt-4o": {
2011
+ "id": "gpt-4o",
2012
+ "name": "GPT-4o",
2013
+ "description": "Omni-era GPT for multimodal chat, practical coding, and general assistants",
2014
+ "family": "gpt",
2015
+ "attachment": true,
2016
+ "reasoning": false,
2017
+ "tool_call": true,
2018
+ "structured_output": true,
2019
+ "temperature": true,
2020
+ "knowledge": "2023-09",
2021
+ "release_date": "2024-05-13",
2022
+ "last_updated": "2024-08-06",
2023
+ "modalities": {
2024
+ "input": [
2025
+ "text",
2026
+ "image",
2027
+ "pdf"
2028
+ ],
2029
+ "output": [
2030
+ "text"
2031
+ ]
2032
+ },
2033
+ "open_weights": false,
2034
+ "limit": {
2035
+ "context": 128000,
2036
+ "output": 16384
2037
+ },
2038
+ "cost": {
2039
+ "input": 2.5,
2040
+ "output": 10,
2041
+ "cache_read": 1.25
2042
+ }
2043
+ },
2044
+ "gpt-5.6-luna": {
2045
+ "id": "gpt-5.6-luna",
2046
+ "name": "GPT-5.6 Luna",
2047
+ "description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
2048
+ "family": "gpt-luna",
2039
2049
  "attachment": true,
2040
2050
  "reasoning": true,
2041
2051
  "reasoning_options": [
2042
2052
  {
2043
2053
  "type": "effort",
2044
2054
  "values": [
2055
+ "none",
2056
+ "low",
2045
2057
  "medium",
2046
2058
  "high",
2047
- "xhigh"
2059
+ "xhigh",
2060
+ "max"
2048
2061
  ]
2049
2062
  }
2050
2063
  ],
2051
2064
  "tool_call": true,
2052
- "structured_output": false,
2065
+ "structured_output": true,
2053
2066
  "temperature": false,
2054
- "knowledge": "2025-08-31",
2055
- "release_date": "2026-03-05",
2056
- "last_updated": "2026-03-05",
2067
+ "knowledge": "2026-02-16",
2068
+ "release_date": "2026-07-09",
2069
+ "last_updated": "2026-07-09",
2057
2070
  "modalities": {
2058
2071
  "input": [
2059
2072
  "text",
2060
- "image"
2073
+ "image",
2074
+ "pdf"
2061
2075
  ],
2062
2076
  "output": [
2063
2077
  "text"
@@ -2069,13 +2083,43 @@
2069
2083
  "input": 922000,
2070
2084
  "output": 128000
2071
2085
  },
2086
+ "experimental": {
2087
+ "modes": {
2088
+ "fast": {
2089
+ "cost": {
2090
+ "input": 2,
2091
+ "output": 12,
2092
+ "cache_read": 0.2,
2093
+ "cache_write": 2.5
2094
+ },
2095
+ "provider": {
2096
+ "body": {
2097
+ "service_tier": "priority"
2098
+ }
2099
+ }
2100
+ },
2101
+ "pro": {
2102
+ "provider": {
2103
+ "body": {
2104
+ "reasoning": {
2105
+ "mode": "pro"
2106
+ }
2107
+ }
2108
+ }
2109
+ }
2110
+ }
2111
+ },
2072
2112
  "cost": {
2073
- "input": 30,
2074
- "output": 180,
2113
+ "input": 1,
2114
+ "output": 6,
2115
+ "cache_read": 0.1,
2116
+ "cache_write": 1.25,
2075
2117
  "tiers": [
2076
2118
  {
2077
- "input": 60,
2078
- "output": 270,
2119
+ "input": 2,
2120
+ "output": 9,
2121
+ "cache_read": 0.2,
2122
+ "cache_write": 2.5,
2079
2123
  "tier": {
2080
2124
  "type": "context",
2081
2125
  "size": 272000
@@ -2083,15 +2127,72 @@
2083
2127
  }
2084
2128
  ],
2085
2129
  "context_over_200k": {
2086
- "input": 60,
2087
- "output": 270
2130
+ "input": 2,
2131
+ "output": 9,
2132
+ "cache_read": 0.2,
2133
+ "cache_write": 2.5
2088
2134
  }
2089
2135
  }
2090
2136
  },
2091
- "gpt-5.5-pro": {
2092
- "id": "gpt-5.5-pro",
2093
- "name": "GPT-5.5 Pro",
2094
- "description": "Highest-accuracy GPT-5.5 tier for slower, precision-heavy reasoning and coding",
2137
+ "chatgpt-image-latest": {
2138
+ "id": "chatgpt-image-latest",
2139
+ "name": "chatgpt-image-latest",
2140
+ "description": "Image model for prompt-driven generation, editing, and visual design workflows",
2141
+ "family": "gpt-image",
2142
+ "attachment": true,
2143
+ "reasoning": false,
2144
+ "tool_call": false,
2145
+ "temperature": false,
2146
+ "release_date": "2025-12-16",
2147
+ "last_updated": "2025-12-16",
2148
+ "modalities": {
2149
+ "input": [
2150
+ "text",
2151
+ "image"
2152
+ ],
2153
+ "output": [
2154
+ "text",
2155
+ "image"
2156
+ ]
2157
+ },
2158
+ "open_weights": false,
2159
+ "limit": {
2160
+ "context": 0,
2161
+ "input": 0,
2162
+ "output": 0
2163
+ }
2164
+ },
2165
+ "gpt-image-1": {
2166
+ "id": "gpt-image-1",
2167
+ "name": "gpt-image-1",
2168
+ "description": "OpenAI image model for production generation, edits, and brand-safe visual workflows",
2169
+ "family": "gpt-image",
2170
+ "attachment": true,
2171
+ "reasoning": false,
2172
+ "tool_call": false,
2173
+ "temperature": false,
2174
+ "release_date": "2025-04-24",
2175
+ "last_updated": "2025-04-24",
2176
+ "modalities": {
2177
+ "input": [
2178
+ "text",
2179
+ "image"
2180
+ ],
2181
+ "output": [
2182
+ "image"
2183
+ ]
2184
+ },
2185
+ "open_weights": false,
2186
+ "limit": {
2187
+ "context": 0,
2188
+ "input": 0,
2189
+ "output": 0
2190
+ }
2191
+ },
2192
+ "gpt-5-pro": {
2193
+ "id": "gpt-5-pro",
2194
+ "name": "GPT-5 Pro",
2195
+ "description": "Higher-accuracy GPT-5 tier for tough analysis, coding reviews, and planning",
2095
2196
  "family": "gpt-pro",
2096
2197
  "attachment": true,
2097
2198
  "reasoning": true,
@@ -2099,23 +2200,20 @@
2099
2200
  {
2100
2201
  "type": "effort",
2101
2202
  "values": [
2102
- "medium",
2103
- "high",
2104
- "xhigh"
2203
+ "high"
2105
2204
  ]
2106
2205
  }
2107
2206
  ],
2108
2207
  "tool_call": true,
2109
2208
  "structured_output": true,
2110
2209
  "temperature": false,
2111
- "knowledge": "2025-12-01",
2112
- "release_date": "2026-04-23",
2113
- "last_updated": "2026-04-23",
2210
+ "knowledge": "2024-09-30",
2211
+ "release_date": "2025-10-06",
2212
+ "last_updated": "2025-10-06",
2114
2213
  "modalities": {
2115
2214
  "input": [
2116
2215
  "text",
2117
- "image",
2118
- "pdf"
2216
+ "image"
2119
2217
  ],
2120
2218
  "output": [
2121
2219
  "text"
@@ -2123,47 +2221,42 @@
2123
2221
  },
2124
2222
  "open_weights": false,
2125
2223
  "limit": {
2126
- "context": 1050000,
2127
- "input": 922000,
2128
- "output": 128000
2224
+ "context": 400000,
2225
+ "input": 272000,
2226
+ "output": 272000
2129
2227
  },
2130
2228
  "cost": {
2131
- "input": 30,
2132
- "output": 180,
2133
- "tiers": [
2134
- {
2135
- "input": 60,
2136
- "output": 270,
2137
- "tier": {
2138
- "type": "context",
2139
- "size": 272000
2140
- }
2141
- }
2142
- ],
2143
- "context_over_200k": {
2144
- "input": 60,
2145
- "output": 270
2146
- }
2229
+ "input": 15,
2230
+ "output": 120
2147
2231
  }
2148
2232
  },
2149
- "gpt-4o-mini": {
2150
- "id": "gpt-4o-mini",
2151
- "name": "GPT-4o mini",
2152
- "description": "Small omni GPT for cheap multimodal assistance and production-scale traffic",
2153
- "family": "gpt-mini",
2233
+ "gpt-5.2-pro": {
2234
+ "id": "gpt-5.2-pro",
2235
+ "name": "GPT-5.2 Pro",
2236
+ "description": "Higher-accuracy GPT-5.2 variant for tougher reasoning and review workflows",
2237
+ "family": "gpt-pro",
2154
2238
  "attachment": true,
2155
- "reasoning": false,
2239
+ "reasoning": true,
2240
+ "reasoning_options": [
2241
+ {
2242
+ "type": "effort",
2243
+ "values": [
2244
+ "medium",
2245
+ "high",
2246
+ "xhigh"
2247
+ ]
2248
+ }
2249
+ ],
2156
2250
  "tool_call": true,
2157
- "structured_output": true,
2158
- "temperature": true,
2159
- "knowledge": "2023-09",
2160
- "release_date": "2024-07-18",
2161
- "last_updated": "2024-07-18",
2251
+ "structured_output": false,
2252
+ "temperature": false,
2253
+ "knowledge": "2025-08-31",
2254
+ "release_date": "2025-12-11",
2255
+ "last_updated": "2025-12-11",
2162
2256
  "modalities": {
2163
2257
  "input": [
2164
2258
  "text",
2165
- "image",
2166
- "pdf"
2259
+ "image"
2167
2260
  ],
2168
2261
  "output": [
2169
2262
  "text"
@@ -2171,20 +2264,20 @@
2171
2264
  },
2172
2265
  "open_weights": false,
2173
2266
  "limit": {
2174
- "context": 128000,
2175
- "output": 16384
2267
+ "context": 400000,
2268
+ "input": 272000,
2269
+ "output": 128000
2176
2270
  },
2177
2271
  "cost": {
2178
- "input": 0.15,
2179
- "output": 0.6,
2180
- "cache_read": 0.075
2272
+ "input": 21,
2273
+ "output": 168
2181
2274
  }
2182
2275
  },
2183
2276
  "gpt-5.6-sol": {
2184
2277
  "id": "gpt-5.6-sol",
2185
2278
  "name": "GPT-5.6 Sol",
2186
2279
  "description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
2187
- "family": "gpt",
2280
+ "family": "gpt-sol",
2188
2281
  "attachment": true,
2189
2282
  "reasoning": true,
2190
2283
  "reasoning_options": [
@@ -2273,140 +2366,17 @@
2273
2366
  }
2274
2367
  }
2275
2368
  },
2276
- "gpt-5-codex": {
2277
- "id": "gpt-5-codex",
2278
- "name": "GPT-5-Codex",
2279
- "description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
2280
- "family": "gpt-codex",
2281
- "attachment": false,
2282
- "reasoning": true,
2283
- "reasoning_options": [
2284
- {
2285
- "type": "effort",
2286
- "values": [
2287
- "low",
2288
- "medium",
2289
- "high"
2290
- ]
2291
- }
2292
- ],
2293
- "tool_call": true,
2294
- "structured_output": true,
2295
- "temperature": false,
2296
- "knowledge": "2024-09-30",
2297
- "release_date": "2025-09-15",
2298
- "last_updated": "2025-09-15",
2299
- "modalities": {
2300
- "input": [
2301
- "text",
2302
- "image"
2303
- ],
2304
- "output": [
2305
- "text"
2306
- ]
2307
- },
2308
- "open_weights": false,
2309
- "limit": {
2310
- "context": 400000,
2311
- "input": 272000,
2312
- "output": 128000
2313
- },
2314
- "cost": {
2315
- "input": 1.25,
2316
- "output": 10,
2317
- "cache_read": 0.125
2318
- }
2319
- },
2320
- "gpt-5.2-codex": {
2321
- "id": "gpt-5.2-codex",
2322
- "name": "GPT-5.2 Codex",
2323
- "description": "Code-specialist GPT for repository edits, reviews, and long-running software agents",
2324
- "family": "gpt-codex",
2325
- "attachment": true,
2326
- "reasoning": true,
2327
- "reasoning_options": [
2328
- {
2329
- "type": "effort",
2330
- "values": [
2331
- "low",
2332
- "medium",
2333
- "high",
2334
- "xhigh"
2335
- ]
2336
- }
2337
- ],
2338
- "tool_call": true,
2339
- "structured_output": true,
2340
- "temperature": false,
2341
- "knowledge": "2025-08-31",
2342
- "release_date": "2025-12-11",
2343
- "last_updated": "2025-12-11",
2344
- "modalities": {
2345
- "input": [
2346
- "text",
2347
- "image",
2348
- "pdf"
2349
- ],
2350
- "output": [
2351
- "text"
2352
- ]
2353
- },
2354
- "open_weights": false,
2355
- "limit": {
2356
- "context": 400000,
2357
- "input": 272000,
2358
- "output": 128000
2359
- },
2360
- "cost": {
2361
- "input": 1.75,
2362
- "output": 14,
2363
- "cache_read": 0.175
2364
- }
2365
- },
2366
- "gpt-image-2": {
2367
- "id": "gpt-image-2",
2368
- "name": "gpt-image-2",
2369
- "description": "Image model for prompt-driven generation, editing, and visual design workflows",
2370
- "family": "gpt-image",
2371
- "attachment": true,
2372
- "reasoning": false,
2373
- "tool_call": false,
2374
- "temperature": false,
2375
- "release_date": "2026-04-21",
2376
- "last_updated": "2026-04-21",
2377
- "modalities": {
2378
- "input": [
2379
- "text",
2380
- "image"
2381
- ],
2382
- "output": [
2383
- "image"
2384
- ]
2385
- },
2386
- "open_weights": false,
2387
- "limit": {
2388
- "context": 0,
2389
- "input": 0,
2390
- "output": 0
2391
- },
2392
- "cost": {
2393
- "input": 5,
2394
- "output": 30,
2395
- "cache_read": 1.25
2396
- }
2397
- },
2398
- "gpt-5.1": {
2399
- "id": "gpt-5.1",
2400
- "name": "GPT-5.1",
2401
- "description": "Sharper GPT-5 generation for coding, product work, and tool-assisted tasks",
2402
- "family": "gpt",
2369
+ "o3-pro": {
2370
+ "id": "o3-pro",
2371
+ "name": "o3-pro",
2372
+ "description": "High-effort o3 tier for difficult technical reasoning and careful answers",
2373
+ "family": "o-pro",
2403
2374
  "attachment": true,
2404
2375
  "reasoning": true,
2405
2376
  "reasoning_options": [
2406
2377
  {
2407
2378
  "type": "effort",
2408
2379
  "values": [
2409
- "none",
2410
2380
  "low",
2411
2381
  "medium",
2412
2382
  "high"
@@ -2416,9 +2386,9 @@
2416
2386
  "tool_call": true,
2417
2387
  "structured_output": true,
2418
2388
  "temperature": false,
2419
- "knowledge": "2024-09-30",
2420
- "release_date": "2025-11-13",
2421
- "last_updated": "2025-11-13",
2389
+ "knowledge": "2024-05",
2390
+ "release_date": "2025-06-10",
2391
+ "last_updated": "2025-06-10",
2422
2392
  "modalities": {
2423
2393
  "input": [
2424
2394
  "text",
@@ -2430,14 +2400,12 @@
2430
2400
  },
2431
2401
  "open_weights": false,
2432
2402
  "limit": {
2433
- "context": 400000,
2434
- "input": 272000,
2435
- "output": 128000
2403
+ "context": 200000,
2404
+ "output": 100000
2436
2405
  },
2437
2406
  "cost": {
2438
- "input": 1.25,
2439
- "output": 10,
2440
- "cache_read": 0.125
2407
+ "input": 20,
2408
+ "output": 80
2441
2409
  }
2442
2410
  },
2443
2411
  "gpt-5.5": {
@@ -2518,6 +2486,38 @@
2518
2486
  "cache_read": 1
2519
2487
  }
2520
2488
  }
2489
+ },
2490
+ "gpt-image-2": {
2491
+ "id": "gpt-image-2",
2492
+ "name": "gpt-image-2",
2493
+ "description": "Image model for prompt-driven generation, editing, and visual design workflows",
2494
+ "family": "gpt-image",
2495
+ "attachment": true,
2496
+ "reasoning": false,
2497
+ "tool_call": false,
2498
+ "temperature": false,
2499
+ "release_date": "2026-04-21",
2500
+ "last_updated": "2026-04-21",
2501
+ "modalities": {
2502
+ "input": [
2503
+ "text",
2504
+ "image"
2505
+ ],
2506
+ "output": [
2507
+ "image"
2508
+ ]
2509
+ },
2510
+ "open_weights": false,
2511
+ "limit": {
2512
+ "context": 0,
2513
+ "input": 0,
2514
+ "output": 0
2515
+ },
2516
+ "cost": {
2517
+ "input": 5,
2518
+ "output": 30,
2519
+ "cache_read": 1.25
2520
+ }
2521
2521
  }
2522
2522
  }
2523
2523
  }