llm.rb 12.4.0 → 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 (40) hide show
  1. checksums.yaml +4 -4
  2. data/CHANGELOG.md +240 -0
  3. data/README.md +38 -8
  4. data/data/anthropic.json +249 -249
  5. data/data/bedrock.json +1964 -1964
  6. data/data/deepinfra.json +589 -589
  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 +1133 -1133
  11. data/data/xai.json +138 -138
  12. data/data/zai.json +165 -165
  13. data/lib/llm/agent.rb +3 -1
  14. data/lib/llm/buffer.rb +15 -0
  15. data/lib/llm/context.rb +4 -5
  16. data/lib/llm/function.rb +2 -35
  17. data/lib/llm/message.rb +12 -2
  18. data/lib/llm/provider.rb +11 -1
  19. data/lib/llm/providers/anthropic/stream_parser.rb +3 -12
  20. data/lib/llm/providers/anthropic.rb +11 -0
  21. data/lib/llm/providers/bedrock/stream_parser.rb +4 -13
  22. data/lib/llm/providers/google/stream_parser.rb +3 -12
  23. data/lib/llm/providers/google.rb +7 -0
  24. data/lib/llm/providers/mistral.rb +11 -0
  25. data/lib/llm/providers/ollama/stream_parser.rb +4 -4
  26. data/lib/llm/providers/ollama.rb +11 -0
  27. data/lib/llm/providers/openai/responses/stream_parser.rb +4 -14
  28. data/lib/llm/providers/openai/responses.rb +10 -0
  29. data/lib/llm/providers/openai/stream_parser.rb +5 -15
  30. data/lib/llm/providers/openai.rb +11 -0
  31. data/lib/llm/repl/command.rb +121 -5
  32. data/lib/llm/repl/commands/exit.rb +4 -0
  33. data/lib/llm/repl/commands/help.rb +24 -0
  34. data/lib/llm/repl/input.rb +4 -1
  35. data/lib/llm/repl/stream.rb +3 -8
  36. data/lib/llm/repl.rb +58 -12
  37. data/lib/llm/stream.rb +2 -7
  38. data/lib/llm/version.rb +1 -1
  39. data/resources/deepdive.md +1 -1
  40. metadata +2 -1
data/data/mistral.json CHANGED
@@ -7,18 +7,18 @@
7
7
  "name": "Mistral",
8
8
  "doc": "https://docs.mistral.ai/getting-started/models/",
9
9
  "models": {
10
- "codestral-latest": {
11
- "id": "codestral-latest",
12
- "name": "Codestral (latest)",
13
- "description": "Mistral code model for completions, refactors, and developer IDE workflows",
14
- "family": "codestral",
10
+ "open-mistral-nemo": {
11
+ "id": "open-mistral-nemo",
12
+ "name": "Open Mistral Nemo",
13
+ "description": "Legacy model retained for compatibility with older integrations",
14
+ "family": "mistral-nemo",
15
15
  "attachment": false,
16
16
  "reasoning": false,
17
17
  "tool_call": true,
18
18
  "temperature": true,
19
- "knowledge": "2024-10",
20
- "release_date": "2024-05-29",
21
- "last_updated": "2025-01-04",
19
+ "knowledge": "2024-07",
20
+ "release_date": "2024-07-01",
21
+ "last_updated": "2024-07-01",
22
22
  "modalities": {
23
23
  "input": [
24
24
  "text"
@@ -29,26 +29,57 @@
29
29
  },
30
30
  "open_weights": true,
31
31
  "limit": {
32
- "context": 256000,
33
- "output": 4096
32
+ "context": 128000,
33
+ "output": 128000
34
34
  },
35
+ "status": "deprecated",
35
36
  "cost": {
36
- "input": 0.3,
37
- "output": 0.9
37
+ "input": 0.15,
38
+ "output": 0.15
38
39
  }
39
40
  },
40
- "mistral-large-latest": {
41
- "id": "mistral-large-latest",
42
- "name": "Mistral Large (latest)",
43
- "description": "Flagship Mistral model for advanced reasoning, coding, and multilingual work",
44
- "family": "mistral-large",
45
- "attachment": true,
41
+ "open-mixtral-8x7b": {
42
+ "id": "open-mixtral-8x7b",
43
+ "name": "Mixtral 8x7B",
44
+ "description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
45
+ "family": "mixtral",
46
+ "attachment": false,
46
47
  "reasoning": false,
47
48
  "tool_call": true,
48
49
  "temperature": true,
49
- "knowledge": "2024-11",
50
- "release_date": "2024-11-01",
51
- "last_updated": "2025-12-02",
50
+ "knowledge": "2024-01",
51
+ "release_date": "2023-12-11",
52
+ "last_updated": "2023-12-11",
53
+ "modalities": {
54
+ "input": [
55
+ "text"
56
+ ],
57
+ "output": [
58
+ "text"
59
+ ]
60
+ },
61
+ "open_weights": true,
62
+ "limit": {
63
+ "context": 32000,
64
+ "output": 32000
65
+ },
66
+ "cost": {
67
+ "input": 0.7,
68
+ "output": 0.7
69
+ }
70
+ },
71
+ "mistral-small-2506": {
72
+ "id": "mistral-small-2506",
73
+ "name": "Mistral Small 3.2",
74
+ "description": "Efficient Mistral model for fast chat, extraction, and production assistants",
75
+ "family": "mistral-small",
76
+ "attachment": false,
77
+ "reasoning": false,
78
+ "tool_call": true,
79
+ "temperature": true,
80
+ "knowledge": "2025-03",
81
+ "release_date": "2025-06-20",
82
+ "last_updated": "2025-06-20",
52
83
  "modalities": {
53
84
  "input": [
54
85
  "text",
@@ -60,26 +91,27 @@
60
91
  },
61
92
  "open_weights": true,
62
93
  "limit": {
63
- "context": 262144,
64
- "output": 262144
94
+ "context": 128000,
95
+ "output": 16384
65
96
  },
66
97
  "cost": {
67
- "input": 0.5,
68
- "output": 1.5
98
+ "input": 0.1,
99
+ "output": 0.3
69
100
  }
70
101
  },
71
- "open-mistral-7b": {
72
- "id": "open-mistral-7b",
73
- "name": "Mistral 7B",
74
- "description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
75
- "family": "mistral",
102
+ "magistral-small": {
103
+ "id": "magistral-small",
104
+ "name": "Magistral Small",
105
+ "description": "Mistral reasoning model for transparent analysis, math, and complex decisions",
106
+ "family": "magistral-small",
76
107
  "attachment": false,
77
- "reasoning": false,
108
+ "reasoning": true,
109
+ "reasoning_options": [],
78
110
  "tool_call": true,
79
111
  "temperature": true,
80
- "knowledge": "2023-12",
81
- "release_date": "2023-09-27",
82
- "last_updated": "2023-09-27",
112
+ "knowledge": "2025-06",
113
+ "release_date": "2025-03-17",
114
+ "last_updated": "2025-03-17",
83
115
  "modalities": {
84
116
  "input": [
85
117
  "text"
@@ -90,12 +122,12 @@
90
122
  },
91
123
  "open_weights": true,
92
124
  "limit": {
93
- "context": 8000,
94
- "output": 8000
125
+ "context": 128000,
126
+ "output": 128000
95
127
  },
96
128
  "cost": {
97
- "input": 0.25,
98
- "output": 0.25
129
+ "input": 0.5,
130
+ "output": 1.5
99
131
  }
100
132
  },
101
133
  "devstral-small-2507": {
@@ -129,21 +161,22 @@
129
161
  "output": 0.3
130
162
  }
131
163
  },
132
- "ministral-3b-latest": {
133
- "id": "ministral-3b-latest",
134
- "name": "Ministral 3B (latest)",
135
- "description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
136
- "family": "ministral",
164
+ "labs-devstral-small-2512": {
165
+ "id": "labs-devstral-small-2512",
166
+ "name": "Devstral Small 2",
167
+ "description": "Legacy model retained for compatibility with older integrations",
168
+ "family": "devstral",
137
169
  "attachment": false,
138
170
  "reasoning": false,
139
171
  "tool_call": true,
140
172
  "temperature": true,
141
- "knowledge": "2024-10",
142
- "release_date": "2024-10-01",
143
- "last_updated": "2024-10-04",
173
+ "knowledge": "2025-12",
174
+ "release_date": "2025-12-09",
175
+ "last_updated": "2025-12-09",
144
176
  "modalities": {
145
177
  "input": [
146
- "text"
178
+ "text",
179
+ "image"
147
180
  ],
148
181
  "output": [
149
182
  "text"
@@ -151,26 +184,36 @@
151
184
  },
152
185
  "open_weights": true,
153
186
  "limit": {
154
- "context": 128000,
155
- "output": 128000
187
+ "context": 256000,
188
+ "output": 256000
156
189
  },
190
+ "status": "deprecated",
157
191
  "cost": {
158
- "input": 0.04,
159
- "output": 0.04
192
+ "input": 0,
193
+ "output": 0
160
194
  }
161
195
  },
162
- "pixtral-large-latest": {
163
- "id": "pixtral-large-latest",
164
- "name": "Pixtral Large (latest)",
165
- "description": "Mistral's larger vision model for document-heavy image understanding and chat",
166
- "family": "pixtral",
196
+ "mistral-medium-latest": {
197
+ "id": "mistral-medium-latest",
198
+ "name": "Mistral Medium (latest)",
199
+ "description": "Balanced Mistral model for enterprise assistants, multilingual work, and tools",
200
+ "family": "mistral-medium",
167
201
  "attachment": true,
168
- "reasoning": false,
202
+ "reasoning": true,
203
+ "reasoning_options": [
204
+ {
205
+ "type": "effort",
206
+ "values": [
207
+ "none",
208
+ "high"
209
+ ]
210
+ }
211
+ ],
169
212
  "tool_call": true,
213
+ "structured_output": true,
170
214
  "temperature": true,
171
- "knowledge": "2024-11",
172
- "release_date": "2024-11-01",
173
- "last_updated": "2024-11-04",
215
+ "release_date": "2026-04-29",
216
+ "last_updated": "2026-04-29",
174
217
  "modalities": {
175
218
  "input": [
176
219
  "text",
@@ -182,26 +225,26 @@
182
225
  },
183
226
  "open_weights": true,
184
227
  "limit": {
185
- "context": 128000,
186
- "output": 128000
228
+ "context": 262144,
229
+ "output": 262144
187
230
  },
188
231
  "cost": {
189
- "input": 2,
190
- "output": 6
232
+ "input": 1.5,
233
+ "output": 7.5
191
234
  }
192
235
  },
193
- "mistral-nemo": {
194
- "id": "mistral-nemo",
195
- "name": "Mistral Nemo",
196
- "description": "Efficient Mistral-NVIDIA open model for multilingual chat and local deployment",
197
- "family": "mistral-nemo",
236
+ "devstral-medium-latest": {
237
+ "id": "devstral-medium-latest",
238
+ "name": "Devstral 2 (latest)",
239
+ "description": "Legacy model retained for compatibility with older integrations",
240
+ "family": "devstral",
198
241
  "attachment": false,
199
242
  "reasoning": false,
200
243
  "tool_call": true,
201
244
  "temperature": true,
202
- "knowledge": "2024-07",
203
- "release_date": "2024-07-01",
204
- "last_updated": "2024-07-01",
245
+ "knowledge": "2025-12",
246
+ "release_date": "2025-12-02",
247
+ "last_updated": "2025-12-02",
205
248
  "modalities": {
206
249
  "input": [
207
250
  "text"
@@ -212,25 +255,27 @@
212
255
  },
213
256
  "open_weights": true,
214
257
  "limit": {
215
- "context": 128000,
216
- "output": 128000
258
+ "context": 262144,
259
+ "output": 262144
217
260
  },
261
+ "status": "deprecated",
218
262
  "cost": {
219
- "input": 0.15,
220
- "output": 0.15
263
+ "input": 0.4,
264
+ "output": 2
221
265
  }
222
266
  },
223
- "mistral-embed": {
224
- "id": "mistral-embed",
225
- "name": "Mistral Embed",
226
- "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
227
- "family": "mistral-embed",
267
+ "open-mixtral-8x22b": {
268
+ "id": "open-mixtral-8x22b",
269
+ "name": "Mixtral 8x22B",
270
+ "description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
271
+ "family": "mixtral",
228
272
  "attachment": false,
229
273
  "reasoning": false,
230
- "tool_call": false,
231
- "temperature": false,
232
- "release_date": "2023-12-11",
233
- "last_updated": "2023-12-11",
274
+ "tool_call": true,
275
+ "temperature": true,
276
+ "knowledge": "2024-04",
277
+ "release_date": "2024-04-17",
278
+ "last_updated": "2024-04-17",
234
279
  "modalities": {
235
280
  "input": [
236
281
  "text"
@@ -239,32 +284,31 @@
239
284
  "text"
240
285
  ]
241
286
  },
242
- "open_weights": false,
287
+ "open_weights": true,
243
288
  "limit": {
244
- "context": 8000,
245
- "output": 3072
289
+ "context": 64000,
290
+ "output": 64000
246
291
  },
247
292
  "cost": {
248
- "input": 0.1,
249
- "output": 0
293
+ "input": 2,
294
+ "output": 6
250
295
  }
251
296
  },
252
- "mistral-small-2506": {
253
- "id": "mistral-small-2506",
254
- "name": "Mistral Small 3.2",
255
- "description": "Efficient Mistral model for fast chat, extraction, and production assistants",
256
- "family": "mistral-small",
297
+ "codestral-latest": {
298
+ "id": "codestral-latest",
299
+ "name": "Codestral (latest)",
300
+ "description": "Mistral code model for completions, refactors, and developer IDE workflows",
301
+ "family": "codestral",
257
302
  "attachment": false,
258
303
  "reasoning": false,
259
304
  "tool_call": true,
260
305
  "temperature": true,
261
- "knowledge": "2025-03",
262
- "release_date": "2025-06-20",
263
- "last_updated": "2025-06-20",
306
+ "knowledge": "2024-10",
307
+ "release_date": "2024-05-29",
308
+ "last_updated": "2025-01-04",
264
309
  "modalities": {
265
310
  "input": [
266
- "text",
267
- "image"
311
+ "text"
268
312
  ],
269
313
  "output": [
270
314
  "text"
@@ -272,12 +316,12 @@
272
316
  },
273
317
  "open_weights": true,
274
318
  "limit": {
275
- "context": 128000,
276
- "output": 16384
319
+ "context": 256000,
320
+ "output": 4096
277
321
  },
278
322
  "cost": {
279
- "input": 0.1,
280
- "output": 0.3
323
+ "input": 0.3,
324
+ "output": 0.9
281
325
  }
282
326
  },
283
327
  "ministral-8b-latest": {
@@ -310,39 +354,9 @@
310
354
  "output": 0.1
311
355
  }
312
356
  },
313
- "open-mixtral-8x22b": {
314
- "id": "open-mixtral-8x22b",
315
- "name": "Mixtral 8x22B",
316
- "description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
317
- "family": "mixtral",
318
- "attachment": false,
319
- "reasoning": false,
320
- "tool_call": true,
321
- "temperature": true,
322
- "knowledge": "2024-04",
323
- "release_date": "2024-04-17",
324
- "last_updated": "2024-04-17",
325
- "modalities": {
326
- "input": [
327
- "text"
328
- ],
329
- "output": [
330
- "text"
331
- ]
332
- },
333
- "open_weights": true,
334
- "limit": {
335
- "context": 64000,
336
- "output": 64000
337
- },
338
- "cost": {
339
- "input": 2,
340
- "output": 6
341
- }
342
- },
343
- "mistral-medium-latest": {
344
- "id": "mistral-medium-latest",
345
- "name": "Mistral Medium (latest)",
357
+ "mistral-medium-2604": {
358
+ "id": "mistral-medium-2604",
359
+ "name": "Mistral Medium 3.5",
346
360
  "description": "Balanced Mistral model for enterprise assistants, multilingual work, and tools",
347
361
  "family": "mistral-medium",
348
362
  "attachment": true,
@@ -380,21 +394,22 @@
380
394
  "output": 7.5
381
395
  }
382
396
  },
383
- "devstral-small-2505": {
384
- "id": "devstral-small-2505",
385
- "name": "Devstral Small 2505",
386
- "description": "Legacy model retained for compatibility with older integrations",
387
- "family": "devstral",
388
- "attachment": false,
397
+ "mistral-large-2512": {
398
+ "id": "mistral-large-2512",
399
+ "name": "Mistral Large 3",
400
+ "description": "Mistral's largest general model for enterprise agents, coding, and multilingual reasoning",
401
+ "family": "mistral-large",
402
+ "attachment": true,
389
403
  "reasoning": false,
390
404
  "tool_call": true,
391
405
  "temperature": true,
392
- "knowledge": "2025-05",
393
- "release_date": "2025-05-07",
394
- "last_updated": "2025-05-07",
406
+ "knowledge": "2024-11",
407
+ "release_date": "2024-11-01",
408
+ "last_updated": "2025-12-02",
395
409
  "modalities": {
396
410
  "input": [
397
- "text"
411
+ "text",
412
+ "image"
398
413
  ],
399
414
  "output": [
400
415
  "text"
@@ -402,31 +417,30 @@
402
417
  },
403
418
  "open_weights": true,
404
419
  "limit": {
405
- "context": 128000,
406
- "output": 128000
420
+ "context": 262144,
421
+ "output": 262144
407
422
  },
408
- "status": "deprecated",
409
423
  "cost": {
410
- "input": 0.1,
411
- "output": 0.3
424
+ "input": 0.5,
425
+ "output": 1.5
412
426
  }
413
427
  },
414
- "magistral-small": {
415
- "id": "magistral-small",
416
- "name": "Magistral Small",
417
- "description": "Mistral reasoning model for transparent analysis, math, and complex decisions",
418
- "family": "magistral-small",
419
- "attachment": false,
420
- "reasoning": true,
421
- "reasoning_options": [],
428
+ "mistral-large-latest": {
429
+ "id": "mistral-large-latest",
430
+ "name": "Mistral Large (latest)",
431
+ "description": "Flagship Mistral model for advanced reasoning, coding, and multilingual work",
432
+ "family": "mistral-large",
433
+ "attachment": true,
434
+ "reasoning": false,
422
435
  "tool_call": true,
423
436
  "temperature": true,
424
- "knowledge": "2025-06",
425
- "release_date": "2025-03-17",
426
- "last_updated": "2025-03-17",
437
+ "knowledge": "2024-11",
438
+ "release_date": "2024-11-01",
439
+ "last_updated": "2025-12-02",
427
440
  "modalities": {
428
441
  "input": [
429
- "text"
442
+ "text",
443
+ "image"
430
444
  ],
431
445
  "output": [
432
446
  "text"
@@ -434,79 +448,59 @@
434
448
  },
435
449
  "open_weights": true,
436
450
  "limit": {
437
- "context": 128000,
438
- "output": 128000
451
+ "context": 262144,
452
+ "output": 262144
439
453
  },
440
454
  "cost": {
441
455
  "input": 0.5,
442
456
  "output": 1.5
443
457
  }
444
458
  },
445
- "mistral-medium-2604": {
446
- "id": "mistral-medium-2604",
447
- "name": "Mistral Medium 3.5",
448
- "description": "Balanced Mistral model for enterprise assistants, multilingual work, and tools",
449
- "family": "mistral-medium",
450
- "attachment": true,
451
- "reasoning": true,
452
- "reasoning_options": [
453
- {
454
- "type": "effort",
455
- "values": [
456
- "none",
457
- "high"
458
- ]
459
- }
460
- ],
461
- "tool_call": true,
462
- "structured_output": true,
463
- "temperature": true,
464
- "release_date": "2026-04-29",
465
- "last_updated": "2026-04-29",
459
+ "mistral-embed": {
460
+ "id": "mistral-embed",
461
+ "name": "Mistral Embed",
462
+ "description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
463
+ "family": "mistral-embed",
464
+ "attachment": false,
465
+ "reasoning": false,
466
+ "tool_call": false,
467
+ "temperature": false,
468
+ "release_date": "2023-12-11",
469
+ "last_updated": "2023-12-11",
466
470
  "modalities": {
467
471
  "input": [
468
- "text",
469
- "image"
472
+ "text"
470
473
  ],
471
474
  "output": [
472
475
  "text"
473
476
  ]
474
477
  },
475
- "open_weights": true,
478
+ "open_weights": false,
476
479
  "limit": {
477
- "context": 262144,
478
- "output": 262144
480
+ "context": 8000,
481
+ "output": 3072
479
482
  },
480
483
  "cost": {
481
- "input": 1.5,
482
- "output": 7.5
484
+ "input": 0.1,
485
+ "output": 0
483
486
  }
484
487
  },
485
- "mistral-small-latest": {
486
- "id": "mistral-small-latest",
487
- "name": "Mistral Small (latest)",
488
- "description": "Efficient Mistral model for fast chat, extraction, and production assistants",
489
- "family": "mistral-small",
490
- "attachment": true,
488
+ "magistral-medium-latest": {
489
+ "id": "magistral-medium-latest",
490
+ "name": "Magistral Medium (latest)",
491
+ "description": "Mistral reasoning model for transparent analysis, math, and complex decisions",
492
+ "family": "magistral-medium",
493
+ "attachment": false,
491
494
  "reasoning": true,
492
- "reasoning_options": [
493
- {
494
- "type": "effort",
495
- "values": [
496
- "none",
497
- "high"
498
- ]
499
- }
500
- ],
495
+ "reasoning_options": [],
501
496
  "tool_call": true,
502
497
  "temperature": true,
503
498
  "knowledge": "2025-06",
504
- "release_date": "2026-03-16",
505
- "last_updated": "2026-03-16",
499
+ "release_date": "2025-03-17",
500
+ "last_updated": "2025-03-20",
506
501
  "modalities": {
507
502
  "input": [
508
- "text",
509
- "image"
503
+ "text"
510
504
  ],
511
505
  "output": [
512
506
  "text"
@@ -514,42 +508,43 @@
514
508
  },
515
509
  "open_weights": true,
516
510
  "limit": {
517
- "context": 256000,
518
- "output": 256000
511
+ "context": 128000,
512
+ "output": 16384
519
513
  },
520
514
  "cost": {
521
- "input": 0.15,
522
- "output": 0.6
515
+ "input": 2,
516
+ "output": 5
523
517
  }
524
518
  },
525
- "open-mixtral-8x7b": {
526
- "id": "open-mixtral-8x7b",
527
- "name": "Mixtral 8x7B",
519
+ "mistral-medium-2508": {
520
+ "id": "mistral-medium-2508",
521
+ "name": "Mistral Medium 3.1",
528
522
  "description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
529
- "family": "mixtral",
530
- "attachment": false,
523
+ "family": "mistral-medium",
524
+ "attachment": true,
531
525
  "reasoning": false,
532
526
  "tool_call": true,
533
527
  "temperature": true,
534
- "knowledge": "2024-01",
535
- "release_date": "2023-12-11",
536
- "last_updated": "2023-12-11",
528
+ "knowledge": "2025-05",
529
+ "release_date": "2025-08-12",
530
+ "last_updated": "2025-08-12",
537
531
  "modalities": {
538
532
  "input": [
539
- "text"
533
+ "text",
534
+ "image"
540
535
  ],
541
536
  "output": [
542
537
  "text"
543
538
  ]
544
539
  },
545
- "open_weights": true,
540
+ "open_weights": false,
546
541
  "limit": {
547
- "context": 32000,
548
- "output": 32000
542
+ "context": 262144,
543
+ "output": 262144
549
544
  },
550
545
  "cost": {
551
- "input": 0.7,
552
- "output": 0.7
546
+ "input": 0.4,
547
+ "output": 2
553
548
  }
554
549
  },
555
550
  "devstral-latest": {
@@ -583,27 +578,18 @@
583
578
  "output": 2
584
579
  }
585
580
  },
586
- "mistral-small-2603": {
587
- "id": "mistral-small-2603",
588
- "name": "Mistral Small 4",
589
- "description": "Fast Mistral production model for chat, extraction, and cost-sensitive agents",
590
- "family": "mistral-small",
581
+ "pixtral-large-latest": {
582
+ "id": "pixtral-large-latest",
583
+ "name": "Pixtral Large (latest)",
584
+ "description": "Mistral's larger vision model for document-heavy image understanding and chat",
585
+ "family": "pixtral",
591
586
  "attachment": true,
592
- "reasoning": true,
593
- "reasoning_options": [
594
- {
595
- "type": "effort",
596
- "values": [
597
- "none",
598
- "high"
599
- ]
600
- }
601
- ],
587
+ "reasoning": false,
602
588
  "tool_call": true,
603
589
  "temperature": true,
604
- "knowledge": "2025-06",
605
- "release_date": "2026-03-16",
606
- "last_updated": "2026-03-16",
590
+ "knowledge": "2024-11",
591
+ "release_date": "2024-11-01",
592
+ "last_updated": "2024-11-04",
607
593
  "modalities": {
608
594
  "input": [
609
595
  "text",
@@ -615,12 +601,12 @@
615
601
  },
616
602
  "open_weights": true,
617
603
  "limit": {
618
- "context": 256000,
619
- "output": 256000
604
+ "context": 128000,
605
+ "output": 128000
620
606
  },
621
607
  "cost": {
622
- "input": 0.15,
623
- "output": 0.6
608
+ "input": 2,
609
+ "output": 6
624
610
  }
625
611
  },
626
612
  "mistral-medium-2505": {
@@ -654,21 +640,31 @@
654
640
  "output": 2
655
641
  }
656
642
  },
657
- "mistral-large-2411": {
658
- "id": "mistral-large-2411",
659
- "name": "Mistral Large 2.1",
660
- "description": "Flagship Mistral model for advanced reasoning, coding, and multilingual work",
661
- "family": "mistral-large",
662
- "attachment": false,
663
- "reasoning": false,
643
+ "mistral-small-latest": {
644
+ "id": "mistral-small-latest",
645
+ "name": "Mistral Small (latest)",
646
+ "description": "Efficient Mistral model for fast chat, extraction, and production assistants",
647
+ "family": "mistral-small",
648
+ "attachment": true,
649
+ "reasoning": true,
650
+ "reasoning_options": [
651
+ {
652
+ "type": "effort",
653
+ "values": [
654
+ "none",
655
+ "high"
656
+ ]
657
+ }
658
+ ],
664
659
  "tool_call": true,
665
660
  "temperature": true,
666
- "knowledge": "2024-11",
667
- "release_date": "2024-11-18",
668
- "last_updated": "2024-11-18",
661
+ "knowledge": "2025-06",
662
+ "release_date": "2026-03-16",
663
+ "last_updated": "2026-03-16",
669
664
  "modalities": {
670
665
  "input": [
671
- "text"
666
+ "text",
667
+ "image"
672
668
  ],
673
669
  "output": [
674
670
  "text"
@@ -676,60 +672,61 @@
676
672
  },
677
673
  "open_weights": true,
678
674
  "limit": {
679
- "context": 131072,
680
- "output": 16384
675
+ "context": 256000,
676
+ "output": 256000
681
677
  },
682
678
  "cost": {
683
- "input": 2,
684
- "output": 6
679
+ "input": 0.15,
680
+ "output": 0.6
685
681
  }
686
682
  },
687
- "mistral-medium-2508": {
688
- "id": "mistral-medium-2508",
689
- "name": "Mistral Medium 3.1",
690
- "description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
691
- "family": "mistral-medium",
692
- "attachment": true,
683
+ "devstral-2512": {
684
+ "id": "devstral-2512",
685
+ "name": "Devstral 2",
686
+ "description": "Mistral's coding-agent model for repository work, terminal tasks, and software fixes",
687
+ "family": "devstral",
688
+ "attachment": false,
693
689
  "reasoning": false,
694
690
  "tool_call": true,
695
691
  "temperature": true,
696
- "knowledge": "2025-05",
697
- "release_date": "2025-08-12",
698
- "last_updated": "2025-08-12",
692
+ "knowledge": "2025-12",
693
+ "release_date": "2025-12-09",
694
+ "last_updated": "2025-12-09",
699
695
  "modalities": {
700
696
  "input": [
701
- "text",
702
- "image"
697
+ "text"
703
698
  ],
704
699
  "output": [
705
700
  "text"
706
701
  ]
707
702
  },
708
- "open_weights": false,
703
+ "open_weights": true,
709
704
  "limit": {
710
705
  "context": 262144,
711
706
  "output": 262144
712
707
  },
708
+ "status": "deprecated",
713
709
  "cost": {
714
710
  "input": 0.4,
715
711
  "output": 2
716
712
  }
717
713
  },
718
- "open-mistral-nemo": {
719
- "id": "open-mistral-nemo",
720
- "name": "Open Mistral Nemo",
721
- "description": "Legacy model retained for compatibility with older integrations",
722
- "family": "mistral-nemo",
723
- "attachment": false,
714
+ "pixtral-12b": {
715
+ "id": "pixtral-12b",
716
+ "name": "Pixtral 12B",
717
+ "description": "Mistral vision-language model for image understanding and multimodal chat",
718
+ "family": "pixtral",
719
+ "attachment": true,
724
720
  "reasoning": false,
725
721
  "tool_call": true,
726
722
  "temperature": true,
727
- "knowledge": "2024-07",
728
- "release_date": "2024-07-01",
729
- "last_updated": "2024-07-01",
723
+ "knowledge": "2024-09",
724
+ "release_date": "2024-09-01",
725
+ "last_updated": "2024-09-01",
730
726
  "modalities": {
731
727
  "input": [
732
- "text"
728
+ "text",
729
+ "image"
733
730
  ],
734
731
  "output": [
735
732
  "text"
@@ -740,28 +737,36 @@
740
737
  "context": 128000,
741
738
  "output": 128000
742
739
  },
743
- "status": "deprecated",
744
740
  "cost": {
745
741
  "input": 0.15,
746
742
  "output": 0.15
747
743
  }
748
744
  },
749
- "magistral-medium-latest": {
750
- "id": "magistral-medium-latest",
751
- "name": "Magistral Medium (latest)",
752
- "description": "Mistral reasoning model for transparent analysis, math, and complex decisions",
753
- "family": "magistral-medium",
754
- "attachment": false,
745
+ "mistral-small-2603": {
746
+ "id": "mistral-small-2603",
747
+ "name": "Mistral Small 4",
748
+ "description": "Fast Mistral production model for chat, extraction, and cost-sensitive agents",
749
+ "family": "mistral-small",
750
+ "attachment": true,
755
751
  "reasoning": true,
756
- "reasoning_options": [],
752
+ "reasoning_options": [
753
+ {
754
+ "type": "effort",
755
+ "values": [
756
+ "none",
757
+ "high"
758
+ ]
759
+ }
760
+ ],
757
761
  "tool_call": true,
758
762
  "temperature": true,
759
763
  "knowledge": "2025-06",
760
- "release_date": "2025-03-17",
761
- "last_updated": "2025-03-20",
764
+ "release_date": "2026-03-16",
765
+ "last_updated": "2026-03-16",
762
766
  "modalities": {
763
767
  "input": [
764
- "text"
768
+ "text",
769
+ "image"
765
770
  ],
766
771
  "output": [
767
772
  "text"
@@ -769,26 +774,26 @@
769
774
  },
770
775
  "open_weights": true,
771
776
  "limit": {
772
- "context": 128000,
773
- "output": 16384
777
+ "context": 256000,
778
+ "output": 256000
774
779
  },
775
780
  "cost": {
776
- "input": 2,
777
- "output": 5
781
+ "input": 0.15,
782
+ "output": 0.6
778
783
  }
779
784
  },
780
- "devstral-medium-latest": {
781
- "id": "devstral-medium-latest",
782
- "name": "Devstral 2 (latest)",
785
+ "devstral-small-2505": {
786
+ "id": "devstral-small-2505",
787
+ "name": "Devstral Small 2505",
783
788
  "description": "Legacy model retained for compatibility with older integrations",
784
789
  "family": "devstral",
785
790
  "attachment": false,
786
791
  "reasoning": false,
787
792
  "tool_call": true,
788
793
  "temperature": true,
789
- "knowledge": "2025-12",
790
- "release_date": "2025-12-02",
791
- "last_updated": "2025-12-02",
794
+ "knowledge": "2025-05",
795
+ "release_date": "2025-05-07",
796
+ "last_updated": "2025-05-07",
792
797
  "modalities": {
793
798
  "input": [
794
799
  "text"
@@ -799,27 +804,27 @@
799
804
  },
800
805
  "open_weights": true,
801
806
  "limit": {
802
- "context": 262144,
803
- "output": 262144
807
+ "context": 128000,
808
+ "output": 128000
804
809
  },
805
810
  "status": "deprecated",
806
811
  "cost": {
807
- "input": 0.4,
808
- "output": 2
812
+ "input": 0.1,
813
+ "output": 0.3
809
814
  }
810
815
  },
811
- "devstral-2512": {
812
- "id": "devstral-2512",
813
- "name": "Devstral 2",
814
- "description": "Mistral's coding-agent model for repository work, terminal tasks, and software fixes",
815
- "family": "devstral",
816
+ "mistral-nemo": {
817
+ "id": "mistral-nemo",
818
+ "name": "Mistral Nemo",
819
+ "description": "Efficient Mistral-NVIDIA open model for multilingual chat and local deployment",
820
+ "family": "mistral-nemo",
816
821
  "attachment": false,
817
822
  "reasoning": false,
818
823
  "tool_call": true,
819
824
  "temperature": true,
820
- "knowledge": "2025-12",
821
- "release_date": "2025-12-09",
822
- "last_updated": "2025-12-09",
825
+ "knowledge": "2024-07",
826
+ "release_date": "2024-07-01",
827
+ "last_updated": "2024-07-01",
823
828
  "modalities": {
824
829
  "input": [
825
830
  "text"
@@ -830,31 +835,29 @@
830
835
  },
831
836
  "open_weights": true,
832
837
  "limit": {
833
- "context": 262144,
834
- "output": 262144
838
+ "context": 128000,
839
+ "output": 128000
835
840
  },
836
- "status": "deprecated",
837
841
  "cost": {
838
- "input": 0.4,
839
- "output": 2
842
+ "input": 0.15,
843
+ "output": 0.15
840
844
  }
841
845
  },
842
- "labs-devstral-small-2512": {
843
- "id": "labs-devstral-small-2512",
844
- "name": "Devstral Small 2",
846
+ "devstral-medium-2507": {
847
+ "id": "devstral-medium-2507",
848
+ "name": "Devstral Medium",
845
849
  "description": "Legacy model retained for compatibility with older integrations",
846
850
  "family": "devstral",
847
851
  "attachment": false,
848
852
  "reasoning": false,
849
853
  "tool_call": true,
850
854
  "temperature": true,
851
- "knowledge": "2025-12",
852
- "release_date": "2025-12-09",
853
- "last_updated": "2025-12-09",
855
+ "knowledge": "2025-05",
856
+ "release_date": "2025-07-10",
857
+ "last_updated": "2025-07-10",
854
858
  "modalities": {
855
859
  "input": [
856
- "text",
857
- "image"
860
+ "text"
858
861
  ],
859
862
  "output": [
860
863
  "text"
@@ -862,31 +865,30 @@
862
865
  },
863
866
  "open_weights": true,
864
867
  "limit": {
865
- "context": 256000,
866
- "output": 256000
868
+ "context": 128000,
869
+ "output": 128000
867
870
  },
868
871
  "status": "deprecated",
869
872
  "cost": {
870
- "input": 0,
871
- "output": 0
873
+ "input": 0.4,
874
+ "output": 2
872
875
  }
873
876
  },
874
- "pixtral-12b": {
875
- "id": "pixtral-12b",
876
- "name": "Pixtral 12B",
877
- "description": "Mistral vision-language model for image understanding and multimodal chat",
878
- "family": "pixtral",
879
- "attachment": true,
877
+ "mistral-large-2411": {
878
+ "id": "mistral-large-2411",
879
+ "name": "Mistral Large 2.1",
880
+ "description": "Flagship Mistral model for advanced reasoning, coding, and multilingual work",
881
+ "family": "mistral-large",
882
+ "attachment": false,
880
883
  "reasoning": false,
881
884
  "tool_call": true,
882
885
  "temperature": true,
883
- "knowledge": "2024-09",
884
- "release_date": "2024-09-01",
885
- "last_updated": "2024-09-01",
886
+ "knowledge": "2024-11",
887
+ "release_date": "2024-11-18",
888
+ "last_updated": "2024-11-18",
886
889
  "modalities": {
887
890
  "input": [
888
- "text",
889
- "image"
891
+ "text"
890
892
  ],
891
893
  "output": [
892
894
  "text"
@@ -894,30 +896,29 @@
894
896
  },
895
897
  "open_weights": true,
896
898
  "limit": {
897
- "context": 128000,
898
- "output": 128000
899
+ "context": 131072,
900
+ "output": 16384
899
901
  },
900
902
  "cost": {
901
- "input": 0.15,
902
- "output": 0.15
903
+ "input": 2,
904
+ "output": 6
903
905
  }
904
906
  },
905
- "mistral-large-2512": {
906
- "id": "mistral-large-2512",
907
- "name": "Mistral Large 3",
908
- "description": "Mistral's largest general model for enterprise agents, coding, and multilingual reasoning",
909
- "family": "mistral-large",
910
- "attachment": true,
907
+ "open-mistral-7b": {
908
+ "id": "open-mistral-7b",
909
+ "name": "Mistral 7B",
910
+ "description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
911
+ "family": "mistral",
912
+ "attachment": false,
911
913
  "reasoning": false,
912
914
  "tool_call": true,
913
915
  "temperature": true,
914
- "knowledge": "2024-11",
915
- "release_date": "2024-11-01",
916
- "last_updated": "2025-12-02",
916
+ "knowledge": "2023-12",
917
+ "release_date": "2023-09-27",
918
+ "last_updated": "2023-09-27",
917
919
  "modalities": {
918
920
  "input": [
919
- "text",
920
- "image"
921
+ "text"
921
922
  ],
922
923
  "output": [
923
924
  "text"
@@ -925,26 +926,26 @@
925
926
  },
926
927
  "open_weights": true,
927
928
  "limit": {
928
- "context": 262144,
929
- "output": 262144
929
+ "context": 8000,
930
+ "output": 8000
930
931
  },
931
932
  "cost": {
932
- "input": 0.5,
933
- "output": 1.5
933
+ "input": 0.25,
934
+ "output": 0.25
934
935
  }
935
936
  },
936
- "devstral-medium-2507": {
937
- "id": "devstral-medium-2507",
938
- "name": "Devstral Medium",
939
- "description": "Legacy model retained for compatibility with older integrations",
940
- "family": "devstral",
937
+ "ministral-3b-latest": {
938
+ "id": "ministral-3b-latest",
939
+ "name": "Ministral 3B (latest)",
940
+ "description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
941
+ "family": "ministral",
941
942
  "attachment": false,
942
943
  "reasoning": false,
943
944
  "tool_call": true,
944
945
  "temperature": true,
945
- "knowledge": "2025-05",
946
- "release_date": "2025-07-10",
947
- "last_updated": "2025-07-10",
946
+ "knowledge": "2024-10",
947
+ "release_date": "2024-10-01",
948
+ "last_updated": "2024-10-04",
948
949
  "modalities": {
949
950
  "input": [
950
951
  "text"
@@ -958,10 +959,9 @@
958
959
  "context": 128000,
959
960
  "output": 128000
960
961
  },
961
- "status": "deprecated",
962
962
  "cost": {
963
- "input": 0.4,
964
- "output": 2
963
+ "input": 0.04,
964
+ "output": 0.04
965
965
  }
966
966
  }
967
967
  }