llm.rb 15.3.0 → 15.4.1

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 (76) hide show
  1. checksums.yaml +4 -4
  2. data/CHANGELOG.md +148 -3
  3. data/README.md +185 -71
  4. data/bin/llm.rb +9 -2
  5. data/data/alibaba.json +47 -2
  6. data/data/anthropic.json +67 -0
  7. data/data/bedrock.json +1828 -425
  8. data/data/deepinfra.json +148 -16
  9. data/data/deepseek.json +7 -4
  10. data/data/google.json +3 -3
  11. data/data/mistral.json +42 -0
  12. data/data/moonshot.json +1 -1
  13. data/data/openai.json +186 -0
  14. data/data/openrouter.json +1783 -440
  15. data/data/xai.json +53 -20
  16. data/data/zai.json +90 -4
  17. data/docs/deepdive/advanced/compaction.md +1 -2
  18. data/docs/deepdive/advanced/context.md +214 -1
  19. data/docs/deepdive/advanced/guard.md +9 -57
  20. data/docs/deepdive/features/builtin_tools.md +14 -16
  21. data/docs/deepdive/features/console.md +5 -0
  22. data/docs/deepdive/features/database.md +85 -10
  23. data/docs/deepdive/fundamentals/agents.md +7 -8
  24. data/docs/deepdive/fundamentals/providers.md +45 -5
  25. data/docs/deepdive/fundamentals/schema.md +73 -0
  26. data/docs/deepdive/fundamentals/tools.md +80 -27
  27. data/docs/deepdive/media/audio.md +8 -19
  28. data/docs/deepdive/media/images.md +8 -10
  29. data/docs/deepdive/media/ocr.md +1 -3
  30. data/docs/deepdive/reference/cost.md +48 -0
  31. data/docs/deepdive/reference/tracer.md +76 -0
  32. data/docs/deepdive.md +1 -1
  33. data/lib/llm/active_record/message.rb +113 -0
  34. data/lib/llm/active_record.rb +1 -0
  35. data/lib/llm/agent.rb +28 -19
  36. data/lib/llm/console/buffer.rb +9 -1
  37. data/lib/llm/console.rb +6 -1
  38. data/lib/llm/context/deserializer.rb +6 -1
  39. data/lib/llm/context.rb +8 -4
  40. data/lib/llm/guard.rb +2 -8
  41. data/lib/llm/provider.rb +16 -0
  42. data/lib/llm/providers/alibaba.rb +15 -0
  43. data/lib/llm/providers/anthropic/error_handler.rb +5 -2
  44. data/lib/llm/providers/anthropic/files.rb +12 -12
  45. data/lib/llm/providers/anthropic/models.rb +2 -2
  46. data/lib/llm/providers/anthropic.rb +2 -2
  47. data/lib/llm/providers/bedrock/error_handler.rb +3 -2
  48. data/lib/llm/providers/bedrock/models.rb +5 -3
  49. data/lib/llm/providers/bedrock.rb +2 -2
  50. data/lib/llm/providers/deepinfra/audio.rb +4 -4
  51. data/lib/llm/providers/deepinfra/images.rb +4 -4
  52. data/lib/llm/providers/google/error_handler.rb +5 -2
  53. data/lib/llm/providers/google/files.rb +10 -10
  54. data/lib/llm/providers/google/images.rb +2 -2
  55. data/lib/llm/providers/google/models.rb +2 -2
  56. data/lib/llm/providers/google.rb +4 -4
  57. data/lib/llm/providers/ollama/error_handler.rb +5 -2
  58. data/lib/llm/providers/ollama/models.rb +2 -2
  59. data/lib/llm/providers/ollama.rb +4 -4
  60. data/lib/llm/providers/openai/audio.rb +6 -6
  61. data/lib/llm/providers/openai/error_handler.rb +5 -2
  62. data/lib/llm/providers/openai/files.rb +10 -10
  63. data/lib/llm/providers/openai/images.rb +4 -4
  64. data/lib/llm/providers/openai/models.rb +2 -2
  65. data/lib/llm/providers/openai/moderations.rb +2 -2
  66. data/lib/llm/providers/openai/responses.rb +6 -6
  67. data/lib/llm/providers/openai/vector_stores.rb +22 -22
  68. data/lib/llm/providers/openai.rb +4 -4
  69. data/lib/llm/providers/xai/images.rb +4 -4
  70. data/lib/llm/tracer/telemetry.rb +4 -4
  71. data/lib/llm/tracer.rb +11 -3
  72. data/lib/llm/transport/execution.rb +8 -4
  73. data/lib/llm/version.rb +1 -1
  74. data/llm.gemspec +0 -6
  75. metadata +3 -5
  76. data/lib/llm/guard/loop.rb +0 -89
data/data/bedrock.json CHANGED
@@ -13,16 +13,21 @@
13
13
  "moonshotai.kimi-k2.5": {
14
14
  "id": "moonshotai.kimi-k2.5",
15
15
  "name": "Kimi K2.5",
16
- "description": "Kimi multimodal agent model for visual understanding, coding, and planning",
17
- "family": "kimi",
18
- "attachment": false,
16
+ "description": "Earlier Kimi frontier model for long-context agents, coding, and multimodal work",
17
+ "family": "kimi-k2",
18
+ "attachment": true,
19
19
  "reasoning": true,
20
- "reasoning_options": [],
20
+ "reasoning_options": [
21
+ {
22
+ "type": "toggle"
23
+ }
24
+ ],
21
25
  "tool_call": true,
22
26
  "interleaved": true,
23
27
  "structured_output": true,
24
28
  "temperature": true,
25
- "release_date": "2026-02-06",
29
+ "knowledge": "2025-01",
30
+ "release_date": "2026-01-27",
26
31
  "last_updated": "2026-02-06",
27
32
  "modalities": {
28
33
  "input": [
@@ -36,7 +41,7 @@
36
41
  "open_weights": true,
37
42
  "limit": {
38
43
  "context": 262143,
39
- "output": 16000
44
+ "output": 16384
40
45
  },
41
46
  "cost": {
42
47
  "input": 0.6,
@@ -51,6 +56,9 @@
51
56
  "attachment": true,
52
57
  "reasoning": true,
53
58
  "reasoning_options": [
59
+ {
60
+ "type": "toggle"
61
+ },
54
62
  {
55
63
  "type": "budget_tokens",
56
64
  "min": 1024
@@ -92,6 +100,9 @@
92
100
  "attachment": true,
93
101
  "reasoning": true,
94
102
  "reasoning_options": [
103
+ {
104
+ "type": "toggle"
105
+ },
95
106
  {
96
107
  "type": "effort",
97
108
  "values": [
@@ -124,10 +135,10 @@
124
135
  "output": 128000
125
136
  },
126
137
  "cost": {
127
- "input": 5,
128
- "output": 25,
129
- "cache_read": 0.5,
130
- "cache_write": 6.25
138
+ "input": 5.5,
139
+ "output": 27.5,
140
+ "cache_read": 0.55,
141
+ "cache_write": 6.875
131
142
  }
132
143
  },
133
144
  "eu.amazon.nova-pro-v1:0": {
@@ -146,7 +157,8 @@
146
157
  "input": [
147
158
  "text",
148
159
  "image",
149
- "video"
160
+ "video",
161
+ "pdf"
150
162
  ],
151
163
  "output": [
152
164
  "text"
@@ -155,25 +167,26 @@
155
167
  "open_weights": false,
156
168
  "limit": {
157
169
  "context": 300000,
158
- "output": 8192
170
+ "output": 10000
159
171
  },
160
172
  "cost": {
161
173
  "input": 0.92,
162
174
  "output": 3.68,
163
- "cache_read": 0.23
175
+ "cache_read": 0.23,
176
+ "cache_write": 0.92
164
177
  }
165
178
  },
166
179
  "us.writer.palmyra-x4-v1:0": {
167
180
  "id": "us.writer.palmyra-x4-v1:0",
168
181
  "name": "Palmyra X4 (US)",
169
- "description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use",
182
+ "description": "Enterprise language model for workflow automation, coding, data analysis, and tool use",
170
183
  "family": "palmyra",
171
184
  "attachment": false,
172
185
  "reasoning": true,
173
186
  "reasoning_options": [],
174
187
  "tool_call": true,
175
188
  "temperature": true,
176
- "release_date": "2025-04-28",
189
+ "release_date": "2024-10-09",
177
190
  "last_updated": "2025-04-28",
178
191
  "modalities": {
179
192
  "input": [
@@ -201,6 +214,9 @@
201
214
  "attachment": true,
202
215
  "reasoning": true,
203
216
  "reasoning_options": [
217
+ {
218
+ "type": "toggle"
219
+ },
204
220
  {
205
221
  "type": "effort",
206
222
  "values": [
@@ -237,10 +253,56 @@
237
253
  "output": 128000
238
254
  },
239
255
  "cost": {
240
- "input": 5,
241
- "output": 25,
242
- "cache_read": 0.5,
243
- "cache_write": 6.25
256
+ "input": 5.5,
257
+ "output": 27.5,
258
+ "cache_read": 0.55,
259
+ "cache_write": 6.875
260
+ }
261
+ },
262
+ "google.gemma-4-31b": {
263
+ "id": "google.gemma-4-31b",
264
+ "name": "Gemma 4 31B IT",
265
+ "description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
266
+ "family": "gemma",
267
+ "attachment": true,
268
+ "reasoning": true,
269
+ "reasoning_options": [
270
+ {
271
+ "type": "effort",
272
+ "values": [
273
+ "none",
274
+ "high"
275
+ ]
276
+ }
277
+ ],
278
+ "tool_call": true,
279
+ "structured_output": true,
280
+ "temperature": true,
281
+ "release_date": "2026-04-02",
282
+ "last_updated": "2026-04-02",
283
+ "modalities": {
284
+ "input": [
285
+ "text",
286
+ "image",
287
+ "video"
288
+ ],
289
+ "output": [
290
+ "text"
291
+ ]
292
+ },
293
+ "open_weights": true,
294
+ "limit": {
295
+ "context": 262144,
296
+ "output": 32768
297
+ },
298
+ "provider": {
299
+ "npm": "@ai-sdk/amazon-bedrock/mantle",
300
+ "api": "https://bedrock-mantle.${AWS_REGION}.api.aws/openai/v1",
301
+ "shape": "responses"
302
+ },
303
+ "cost": {
304
+ "input": 0.14,
305
+ "output": 0.4
244
306
  }
245
307
  },
246
308
  "us.xai.grok-4.6": {
@@ -320,15 +382,15 @@
320
382
  "qwen.qwen3-coder-next": {
321
383
  "id": "qwen.qwen3-coder-next",
322
384
  "name": "Qwen3 Coder Next",
323
- "description": "Qwen coding model for software agents, repository edits, and code reasoning",
385
+ "description": "Open-weight Qwen coding model for agents, repository edits, and multi-turn tool use",
324
386
  "family": "qwen",
325
387
  "attachment": false,
326
- "reasoning": true,
327
- "reasoning_options": [],
388
+ "reasoning": false,
328
389
  "tool_call": true,
329
390
  "structured_output": true,
330
391
  "temperature": true,
331
- "release_date": "2026-02-06",
392
+ "knowledge": "2025-09",
393
+ "release_date": "2026-02-03",
332
394
  "last_updated": "2026-02-06",
333
395
  "modalities": {
334
396
  "input": [
@@ -340,12 +402,12 @@
340
402
  },
341
403
  "open_weights": true,
342
404
  "limit": {
343
- "context": 131072,
405
+ "context": 262144,
344
406
  "output": 65536
345
407
  },
346
408
  "cost": {
347
- "input": 0.22,
348
- "output": 1.8
409
+ "input": 0.5,
410
+ "output": 1.2
349
411
  }
350
412
  },
351
413
  "global.openai.gpt-5.6-luna": {
@@ -369,7 +431,7 @@
369
431
  }
370
432
  ],
371
433
  "tool_call": true,
372
- "structured_output": false,
434
+ "structured_output": true,
373
435
  "temperature": false,
374
436
  "knowledge": "2026-02-16",
375
437
  "release_date": "2026-07-09",
@@ -377,8 +439,7 @@
377
439
  "modalities": {
378
440
  "input": [
379
441
  "text",
380
- "image",
381
- "pdf"
442
+ "image"
382
443
  ],
383
444
  "output": [
384
445
  "text"
@@ -423,6 +484,9 @@
423
484
  "attachment": true,
424
485
  "reasoning": true,
425
486
  "reasoning_options": [
487
+ {
488
+ "type": "toggle"
489
+ },
426
490
  {
427
491
  "type": "effort",
428
492
  "values": [
@@ -502,7 +566,7 @@
502
566
  },
503
567
  "open_weights": false,
504
568
  "limit": {
505
- "context": 272000,
569
+ "context": 1000000,
506
570
  "output": 128000
507
571
  },
508
572
  "provider": {
@@ -516,18 +580,58 @@
516
580
  "cache_read": 0.55
517
581
  }
518
582
  },
583
+ "us-gov.openai.gpt-oss-20b-1:0": {
584
+ "id": "us-gov.openai.gpt-oss-20b-1:0",
585
+ "name": "gpt-oss-20b (GovCloud)",
586
+ "description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
587
+ "family": "gpt-oss",
588
+ "attachment": false,
589
+ "reasoning": true,
590
+ "reasoning_options": [
591
+ {
592
+ "type": "effort",
593
+ "values": [
594
+ "low",
595
+ "medium",
596
+ "high"
597
+ ]
598
+ }
599
+ ],
600
+ "tool_call": true,
601
+ "structured_output": true,
602
+ "temperature": true,
603
+ "release_date": "2025-08-05",
604
+ "last_updated": "2025-08-05",
605
+ "modalities": {
606
+ "input": [
607
+ "text"
608
+ ],
609
+ "output": [
610
+ "text"
611
+ ]
612
+ },
613
+ "open_weights": true,
614
+ "limit": {
615
+ "context": 128000,
616
+ "output": 16384
617
+ },
618
+ "cost": {
619
+ "input": 0.084,
620
+ "output": 0.36
621
+ }
622
+ },
519
623
  "qwen.qwen3-coder-30b-a3b-v1:0": {
520
624
  "id": "qwen.qwen3-coder-30b-a3b-v1:0",
521
- "name": "Qwen3 Coder 30B A3B Instruct",
522
- "description": "Qwen coding model for software agents, repository edits, and code reasoning",
625
+ "name": "Qwen3-Coder 30B-A3B Instruct",
626
+ "description": "Smaller Qwen coder for efficient local agents and repo-level fixes",
523
627
  "family": "qwen",
524
628
  "attachment": false,
525
629
  "reasoning": false,
526
630
  "tool_call": true,
527
631
  "structured_output": true,
528
632
  "temperature": true,
529
- "knowledge": "2024-04",
530
- "release_date": "2025-09-18",
633
+ "knowledge": "2025-04",
634
+ "release_date": "2025-07-31",
531
635
  "last_updated": "2025-09-18",
532
636
  "modalities": {
533
637
  "input": [
@@ -537,7 +641,7 @@
537
641
  "text"
538
642
  ]
539
643
  },
540
- "open_weights": false,
644
+ "open_weights": true,
541
645
  "limit": {
542
646
  "context": 262144,
543
647
  "output": 131072
@@ -555,6 +659,9 @@
555
659
  "attachment": true,
556
660
  "reasoning": true,
557
661
  "reasoning_options": [
662
+ {
663
+ "type": "toggle"
664
+ },
558
665
  {
559
666
  "type": "budget_tokens",
560
667
  "min": 1024
@@ -590,16 +697,15 @@
590
697
  },
591
698
  "qwen.qwen3-235b-a22b-2507-v1:0": {
592
699
  "id": "qwen.qwen3-235b-a22b-2507-v1:0",
593
- "name": "Qwen3 235B A22B 2507",
594
- "description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
700
+ "name": "Qwen3 235B-A22B Instruct 2507",
701
+ "description": "Updated large open Qwen3 MoE instruct model for multilingual chat, coding, and tool use",
595
702
  "family": "qwen",
596
703
  "attachment": false,
597
704
  "reasoning": false,
598
705
  "tool_call": true,
599
706
  "structured_output": true,
600
707
  "temperature": true,
601
- "knowledge": "2024-04",
602
- "release_date": "2025-09-18",
708
+ "release_date": "2025-07-21",
603
709
  "last_updated": "2025-09-18",
604
710
  "modalities": {
605
711
  "input": [
@@ -622,9 +728,9 @@
622
728
  "mistral.ministral-3-3b-instruct": {
623
729
  "id": "mistral.ministral-3-3b-instruct",
624
730
  "name": "Ministral 3 3B",
625
- "description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
731
+ "description": "Compact open vision-language model for edge deployment, instruction following, and tool use",
626
732
  "family": "ministral",
627
- "attachment": false,
733
+ "attachment": true,
628
734
  "reasoning": false,
629
735
  "tool_call": true,
630
736
  "structured_output": true,
@@ -650,6 +756,46 @@
650
756
  "output": 0.1
651
757
  }
652
758
  },
759
+ "us-gov.openai.gpt-oss-120b-1:0": {
760
+ "id": "us-gov.openai.gpt-oss-120b-1:0",
761
+ "name": "gpt-oss-120b (GovCloud)",
762
+ "description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
763
+ "family": "gpt-oss",
764
+ "attachment": false,
765
+ "reasoning": true,
766
+ "reasoning_options": [
767
+ {
768
+ "type": "effort",
769
+ "values": [
770
+ "low",
771
+ "medium",
772
+ "high"
773
+ ]
774
+ }
775
+ ],
776
+ "tool_call": true,
777
+ "structured_output": true,
778
+ "temperature": true,
779
+ "release_date": "2025-08-05",
780
+ "last_updated": "2025-08-05",
781
+ "modalities": {
782
+ "input": [
783
+ "text"
784
+ ],
785
+ "output": [
786
+ "text"
787
+ ]
788
+ },
789
+ "open_weights": true,
790
+ "limit": {
791
+ "context": 128000,
792
+ "output": 16384
793
+ },
794
+ "cost": {
795
+ "input": 0.18,
796
+ "output": 0.72
797
+ }
798
+ },
653
799
  "global.anthropic.claude-sonnet-4-6": {
654
800
  "id": "global.anthropic.claude-sonnet-4-6",
655
801
  "name": "Claude Sonnet 4.6 (Global)",
@@ -658,6 +804,9 @@
658
804
  "attachment": true,
659
805
  "reasoning": true,
660
806
  "reasoning_options": [
807
+ {
808
+ "type": "toggle"
809
+ },
661
810
  {
662
811
  "type": "effort",
663
812
  "values": [
@@ -691,7 +840,7 @@
691
840
  "open_weights": false,
692
841
  "limit": {
693
842
  "context": 1000000,
694
- "output": 64000
843
+ "output": 128000
695
844
  },
696
845
  "cost": {
697
846
  "input": 3,
@@ -737,7 +886,7 @@
737
886
  },
738
887
  "open_weights": false,
739
888
  "limit": {
740
- "context": 272000,
889
+ "context": 1000000,
741
890
  "output": 128000
742
891
  },
743
892
  "provider": {
@@ -755,8 +904,8 @@
755
904
  "id": "mistral.pixtral-large-2502-v1:0",
756
905
  "name": "Pixtral Large (25.02)",
757
906
  "description": "Mistral vision-language model for image understanding and multimodal chat",
758
- "family": "mistral",
759
- "attachment": false,
907
+ "family": "pixtral",
908
+ "attachment": true,
760
909
  "reasoning": false,
761
910
  "tool_call": true,
762
911
  "temperature": true,
@@ -784,13 +933,14 @@
784
933
  "mistral.mistral-large-3-675b-instruct": {
785
934
  "id": "mistral.mistral-large-3-675b-instruct",
786
935
  "name": "Mistral Large 3",
787
- "description": "Flagship Mistral model for advanced reasoning, coding, and multilingual work",
788
- "family": "mistral",
789
- "attachment": false,
936
+ "description": "Mistral's largest general model for enterprise agents, coding, and multilingual reasoning",
937
+ "family": "mistral-large",
938
+ "attachment": true,
790
939
  "reasoning": false,
791
940
  "tool_call": true,
792
941
  "structured_output": true,
793
942
  "temperature": true,
943
+ "knowledge": "2024-11",
794
944
  "release_date": "2025-12-02",
795
945
  "last_updated": "2025-12-02",
796
946
  "modalities": {
@@ -820,6 +970,9 @@
820
970
  "attachment": true,
821
971
  "reasoning": true,
822
972
  "reasoning_options": [
973
+ {
974
+ "type": "toggle"
975
+ },
823
976
  {
824
977
  "type": "effort",
825
978
  "values": [
@@ -836,9 +989,9 @@
836
989
  "tool_call": true,
837
990
  "structured_output": true,
838
991
  "temperature": true,
839
- "knowledge": "2025-03-31",
840
- "release_date": "2025-11-24",
841
- "last_updated": "2025-08-01",
992
+ "knowledge": "2025-05",
993
+ "release_date": "2025-11-01",
994
+ "last_updated": "2025-11-01",
842
995
  "modalities": {
843
996
  "input": [
844
997
  "text",
@@ -884,12 +1037,13 @@
884
1037
  "open_weights": false,
885
1038
  "limit": {
886
1039
  "context": 128000,
887
- "output": 8192
1040
+ "output": 10000
888
1041
  },
889
1042
  "cost": {
890
1043
  "input": 0.035,
891
1044
  "output": 0.14,
892
- "cache_read": 0.00875
1045
+ "cache_read": 0.00875,
1046
+ "cache_write": 0.035
893
1047
  }
894
1048
  },
895
1049
  "jp.anthropic.claude-opus-4-7": {
@@ -900,6 +1054,9 @@
900
1054
  "attachment": true,
901
1055
  "reasoning": true,
902
1056
  "reasoning_options": [
1057
+ {
1058
+ "type": "toggle"
1059
+ },
903
1060
  {
904
1061
  "type": "effort",
905
1062
  "values": [
@@ -932,10 +1089,10 @@
932
1089
  "output": 128000
933
1090
  },
934
1091
  "cost": {
935
- "input": 5,
936
- "output": 25,
937
- "cache_read": 0.5,
938
- "cache_write": 6.25
1092
+ "input": 5.5,
1093
+ "output": 27.5,
1094
+ "cache_read": 0.55,
1095
+ "cache_write": 6.875
939
1096
  }
940
1097
  },
941
1098
  "eu.anthropic.claude-sonnet-5": {
@@ -961,7 +1118,7 @@
961
1118
  }
962
1119
  ],
963
1120
  "tool_call": true,
964
- "structured_output": true,
1121
+ "structured_output": false,
965
1122
  "temperature": false,
966
1123
  "knowledge": "2026-01-31",
967
1124
  "release_date": "2026-06-30",
@@ -1011,12 +1168,13 @@
1011
1168
  "open_weights": false,
1012
1169
  "limit": {
1013
1170
  "context": 128000,
1014
- "output": 8192
1171
+ "output": 10000
1015
1172
  },
1016
1173
  "cost": {
1017
1174
  "input": 0.037,
1018
1175
  "output": 0.148,
1019
- "cache_read": 0.00925
1176
+ "cache_read": 0.00925,
1177
+ "cache_write": 0.037
1020
1178
  }
1021
1179
  },
1022
1180
  "nvidia.nemotron-nano-9b-v2": {
@@ -1029,8 +1187,8 @@
1029
1187
  "tool_call": true,
1030
1188
  "structured_output": true,
1031
1189
  "temperature": true,
1032
- "release_date": "2024-12-01",
1033
- "last_updated": "2024-12-01",
1190
+ "release_date": "2025-08-18",
1191
+ "last_updated": "2025-08-18",
1034
1192
  "modalities": {
1035
1193
  "input": [
1036
1194
  "text"
@@ -1039,10 +1197,10 @@
1039
1197
  "text"
1040
1198
  ]
1041
1199
  },
1042
- "open_weights": false,
1200
+ "open_weights": true,
1043
1201
  "limit": {
1044
- "context": 128000,
1045
- "output": 4096
1202
+ "context": 131072,
1203
+ "output": 8192
1046
1204
  },
1047
1205
  "cost": {
1048
1206
  "input": 0.06,
@@ -1052,11 +1210,14 @@
1052
1210
  "au.anthropic.claude-sonnet-4-6": {
1053
1211
  "id": "au.anthropic.claude-sonnet-4-6",
1054
1212
  "name": "AU Anthropic Claude Sonnet 4.6",
1055
- "description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
1213
+ "description": "Claude workhorse for coding agents, careful analysis, and production cost control",
1056
1214
  "family": "claude-sonnet",
1057
1215
  "attachment": true,
1058
1216
  "reasoning": true,
1059
1217
  "reasoning_options": [
1218
+ {
1219
+ "type": "toggle"
1220
+ },
1060
1221
  {
1061
1222
  "type": "effort",
1062
1223
  "values": [
@@ -1074,9 +1235,9 @@
1074
1235
  "tool_call": true,
1075
1236
  "structured_output": true,
1076
1237
  "temperature": true,
1077
- "knowledge": "2025-08",
1238
+ "knowledge": "2025-08-31",
1078
1239
  "release_date": "2026-02-17",
1079
- "last_updated": "2026-02-17",
1240
+ "last_updated": "2026-03-13",
1080
1241
  "modalities": {
1081
1242
  "input": [
1082
1243
  "text",
@@ -1102,11 +1263,14 @@
1102
1263
  "anthropic.claude-opus-4-7": {
1103
1264
  "id": "anthropic.claude-opus-4-7",
1104
1265
  "name": "Claude Opus 4.7",
1105
- "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
1266
+ "description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
1106
1267
  "family": "claude-opus",
1107
1268
  "attachment": true,
1108
1269
  "reasoning": true,
1109
1270
  "reasoning_options": [
1271
+ {
1272
+ "type": "toggle"
1273
+ },
1110
1274
  {
1111
1275
  "type": "effort",
1112
1276
  "values": [
@@ -1148,24 +1312,25 @@
1148
1312
  "mistral.ministral-3-8b-instruct": {
1149
1313
  "id": "mistral.ministral-3-8b-instruct",
1150
1314
  "name": "Ministral 3 8B",
1151
- "description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
1315
+ "description": "Compact open vision-language model for edge deployment, instruction following, and tool use",
1152
1316
  "family": "ministral",
1153
- "attachment": false,
1317
+ "attachment": true,
1154
1318
  "reasoning": false,
1155
1319
  "tool_call": true,
1156
1320
  "structured_output": true,
1157
1321
  "temperature": true,
1158
- "release_date": "2024-12-01",
1159
- "last_updated": "2024-12-01",
1322
+ "release_date": "2025-12-02",
1323
+ "last_updated": "2025-12-02",
1160
1324
  "modalities": {
1161
1325
  "input": [
1162
- "text"
1326
+ "text",
1327
+ "image"
1163
1328
  ],
1164
1329
  "output": [
1165
1330
  "text"
1166
1331
  ]
1167
1332
  },
1168
- "open_weights": false,
1333
+ "open_weights": true,
1169
1334
  "limit": {
1170
1335
  "context": 128000,
1171
1336
  "output": 4096
@@ -1183,6 +1348,9 @@
1183
1348
  "attachment": true,
1184
1349
  "reasoning": true,
1185
1350
  "reasoning_options": [
1351
+ {
1352
+ "type": "toggle"
1353
+ },
1186
1354
  {
1187
1355
  "type": "budget_tokens",
1188
1356
  "min": 1024
@@ -1210,10 +1378,10 @@
1210
1378
  "output": 64000
1211
1379
  },
1212
1380
  "cost": {
1213
- "input": 3,
1214
- "output": 15,
1215
- "cache_read": 0.3,
1216
- "cache_write": 3.75
1381
+ "input": 3.3,
1382
+ "output": 16.5,
1383
+ "cache_read": 0.33,
1384
+ "cache_write": 4.125
1217
1385
  }
1218
1386
  },
1219
1387
  "us.openai.gpt-5.6-sol": {
@@ -1245,8 +1413,7 @@
1245
1413
  "modalities": {
1246
1414
  "input": [
1247
1415
  "text",
1248
- "image",
1249
- "pdf"
1416
+ "image"
1250
1417
  ],
1251
1418
  "output": [
1252
1419
  "text"
@@ -1299,7 +1466,8 @@
1299
1466
  "input": [
1300
1467
  "text",
1301
1468
  "image",
1302
- "video"
1469
+ "video",
1470
+ "pdf"
1303
1471
  ],
1304
1472
  "output": [
1305
1473
  "text"
@@ -1308,12 +1476,13 @@
1308
1476
  "open_weights": false,
1309
1477
  "limit": {
1310
1478
  "context": 300000,
1311
- "output": 8192
1479
+ "output": 10000
1312
1480
  },
1313
1481
  "cost": {
1314
1482
  "input": 0.069,
1315
1483
  "output": 0.276,
1316
- "cache_read": 0.01725
1484
+ "cache_read": 0.01725,
1485
+ "cache_write": 0.069
1317
1486
  }
1318
1487
  },
1319
1488
  "anthropic.claude-opus-5": {
@@ -1325,7 +1494,10 @@
1325
1494
  "reasoning": true,
1326
1495
  "reasoning_options": [
1327
1496
  {
1328
- "type": "effort",
1497
+ "type": "toggle"
1498
+ },
1499
+ {
1500
+ "type": "effort",
1329
1501
  "values": [
1330
1502
  "low",
1331
1503
  "medium",
@@ -1370,6 +1542,9 @@
1370
1542
  "attachment": true,
1371
1543
  "reasoning": true,
1372
1544
  "reasoning_options": [
1545
+ {
1546
+ "type": "toggle"
1547
+ },
1373
1548
  {
1374
1549
  "type": "effort",
1375
1550
  "values": [
@@ -1412,6 +1587,41 @@
1412
1587
  "cache_write": 6.875
1413
1588
  }
1414
1589
  },
1590
+ "global.moonshotai.kimi-k3": {
1591
+ "id": "global.moonshotai.kimi-k3",
1592
+ "name": "Kimi K3 (Global)",
1593
+ "description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
1594
+ "family": "kimi-k3",
1595
+ "attachment": true,
1596
+ "reasoning": true,
1597
+ "reasoning_options": [],
1598
+ "tool_call": true,
1599
+ "interleaved": true,
1600
+ "structured_output": true,
1601
+ "temperature": false,
1602
+ "release_date": "2026-07-16",
1603
+ "last_updated": "2026-07-16",
1604
+ "modalities": {
1605
+ "input": [
1606
+ "text",
1607
+ "image"
1608
+ ],
1609
+ "output": [
1610
+ "text"
1611
+ ]
1612
+ },
1613
+ "open_weights": true,
1614
+ "limit": {
1615
+ "context": 1048576,
1616
+ "output": 131072
1617
+ },
1618
+ "cost": {
1619
+ "input": 3,
1620
+ "output": 15,
1621
+ "cache_read": 0.3,
1622
+ "cache_write": 3.75
1623
+ }
1624
+ },
1415
1625
  "apac.amazon.nova-pro-v1:0": {
1416
1626
  "id": "apac.amazon.nova-pro-v1:0",
1417
1627
  "name": "Nova Pro (APAC)",
@@ -1428,7 +1638,8 @@
1428
1638
  "input": [
1429
1639
  "text",
1430
1640
  "image",
1431
- "video"
1641
+ "video",
1642
+ "pdf"
1432
1643
  ],
1433
1644
  "output": [
1434
1645
  "text"
@@ -1437,12 +1648,13 @@
1437
1648
  "open_weights": false,
1438
1649
  "limit": {
1439
1650
  "context": 300000,
1440
- "output": 8192
1651
+ "output": 10000
1441
1652
  },
1442
1653
  "cost": {
1443
1654
  "input": 0.84,
1444
1655
  "output": 3.36,
1445
- "cache_read": 0.21
1656
+ "cache_read": 0.21,
1657
+ "cache_write": 0.84
1446
1658
  }
1447
1659
  },
1448
1660
  "anthropic.claude-sonnet-4-6": {
@@ -1453,6 +1665,9 @@
1453
1665
  "attachment": true,
1454
1666
  "reasoning": true,
1455
1667
  "reasoning_options": [
1668
+ {
1669
+ "type": "toggle"
1670
+ },
1456
1671
  {
1457
1672
  "type": "effort",
1458
1673
  "values": [
@@ -1486,13 +1701,48 @@
1486
1701
  "open_weights": false,
1487
1702
  "limit": {
1488
1703
  "context": 1000000,
1489
- "output": 64000
1704
+ "output": 128000
1490
1705
  },
1491
1706
  "cost": {
1492
- "input": 3,
1493
- "output": 15,
1494
- "cache_read": 0.3,
1495
- "cache_write": 3.75
1707
+ "input": 3.3,
1708
+ "output": 16.5,
1709
+ "cache_read": 0.33,
1710
+ "cache_write": 4.125
1711
+ }
1712
+ },
1713
+ "us.moonshotai.kimi-k3": {
1714
+ "id": "us.moonshotai.kimi-k3",
1715
+ "name": "Kimi K3 (US)",
1716
+ "description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
1717
+ "family": "kimi-k3",
1718
+ "attachment": true,
1719
+ "reasoning": true,
1720
+ "reasoning_options": [],
1721
+ "tool_call": true,
1722
+ "interleaved": true,
1723
+ "structured_output": true,
1724
+ "temperature": false,
1725
+ "release_date": "2026-07-16",
1726
+ "last_updated": "2026-07-16",
1727
+ "modalities": {
1728
+ "input": [
1729
+ "text",
1730
+ "image"
1731
+ ],
1732
+ "output": [
1733
+ "text"
1734
+ ]
1735
+ },
1736
+ "open_weights": true,
1737
+ "limit": {
1738
+ "context": 1048576,
1739
+ "output": 131072
1740
+ },
1741
+ "cost": {
1742
+ "input": 3.3,
1743
+ "output": 16.5,
1744
+ "cache_read": 0.33,
1745
+ "cache_write": 4.125
1496
1746
  }
1497
1747
  },
1498
1748
  "apac.amazon.nova-lite-v1:0": {
@@ -1511,7 +1761,8 @@
1511
1761
  "input": [
1512
1762
  "text",
1513
1763
  "image",
1514
- "video"
1764
+ "video",
1765
+ "pdf"
1515
1766
  ],
1516
1767
  "output": [
1517
1768
  "text"
@@ -1520,38 +1771,39 @@
1520
1771
  "open_weights": false,
1521
1772
  "limit": {
1522
1773
  "context": 300000,
1523
- "output": 8192
1774
+ "output": 10000
1524
1775
  },
1525
1776
  "cost": {
1526
1777
  "input": 0.063,
1527
1778
  "output": 0.252,
1528
- "cache_read": 0.01575
1779
+ "cache_read": 0.01575,
1780
+ "cache_write": 0.063
1529
1781
  }
1530
1782
  },
1531
1783
  "mistral.voxtral-mini-3b-2507": {
1532
1784
  "id": "mistral.voxtral-mini-3b-2507",
1533
1785
  "name": "Voxtral Mini 3B 2507",
1534
- "description": "Efficient Mistral model for fast chat, extraction, and production assistants",
1535
- "family": "mistral",
1536
- "attachment": false,
1786
+ "description": "Open audio-language model for speech transcription, audio understanding, and voice-driven tool use",
1787
+ "family": "voxtral",
1788
+ "attachment": true,
1537
1789
  "reasoning": false,
1538
1790
  "tool_call": true,
1539
1791
  "structured_output": true,
1540
1792
  "temperature": true,
1541
- "release_date": "2024-12-01",
1542
- "last_updated": "2024-12-01",
1793
+ "release_date": "2025-07-15",
1794
+ "last_updated": "2025-07-15",
1543
1795
  "modalities": {
1544
1796
  "input": [
1545
- "audio",
1546
- "text"
1797
+ "text",
1798
+ "audio"
1547
1799
  ],
1548
1800
  "output": [
1549
1801
  "text"
1550
1802
  ]
1551
1803
  },
1552
- "open_weights": false,
1804
+ "open_weights": true,
1553
1805
  "limit": {
1554
- "context": 128000,
1806
+ "context": 32768,
1555
1807
  "output": 4096
1556
1808
  },
1557
1809
  "cost": {
@@ -1559,18 +1811,64 @@
1559
1811
  "output": 0.04
1560
1812
  }
1561
1813
  },
1814
+ "google.gemma-4-26b-a4b": {
1815
+ "id": "google.gemma-4-26b-a4b",
1816
+ "name": "Gemma 4 26B A4B IT",
1817
+ "description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
1818
+ "family": "gemma",
1819
+ "attachment": true,
1820
+ "reasoning": true,
1821
+ "reasoning_options": [
1822
+ {
1823
+ "type": "effort",
1824
+ "values": [
1825
+ "none",
1826
+ "high"
1827
+ ]
1828
+ }
1829
+ ],
1830
+ "tool_call": true,
1831
+ "structured_output": true,
1832
+ "temperature": true,
1833
+ "release_date": "2026-04-02",
1834
+ "last_updated": "2026-04-02",
1835
+ "modalities": {
1836
+ "input": [
1837
+ "text",
1838
+ "image",
1839
+ "video"
1840
+ ],
1841
+ "output": [
1842
+ "text"
1843
+ ]
1844
+ },
1845
+ "open_weights": true,
1846
+ "limit": {
1847
+ "context": 262144,
1848
+ "output": 32768
1849
+ },
1850
+ "provider": {
1851
+ "npm": "@ai-sdk/amazon-bedrock/mantle",
1852
+ "api": "https://bedrock-mantle.${AWS_REGION}.api.aws/openai/v1",
1853
+ "shape": "responses"
1854
+ },
1855
+ "cost": {
1856
+ "input": 0.13,
1857
+ "output": 0.4
1858
+ }
1859
+ },
1562
1860
  "nvidia.nemotron-nano-12b-v2": {
1563
1861
  "id": "nvidia.nemotron-nano-12b-v2",
1564
1862
  "name": "NVIDIA Nemotron Nano 12B v2 VL BF16",
1565
1863
  "description": "Nemotron multimodal model for visual reasoning and agentic AI workflows",
1566
1864
  "family": "nemotron",
1567
- "attachment": false,
1865
+ "attachment": true,
1568
1866
  "reasoning": false,
1569
1867
  "tool_call": true,
1570
1868
  "structured_output": true,
1571
1869
  "temperature": true,
1572
- "release_date": "2024-12-01",
1573
- "last_updated": "2024-12-01",
1870
+ "release_date": "2025-10-28",
1871
+ "last_updated": "2025-10-28",
1574
1872
  "modalities": {
1575
1873
  "input": [
1576
1874
  "text",
@@ -1580,10 +1878,10 @@
1580
1878
  "text"
1581
1879
  ]
1582
1880
  },
1583
- "open_weights": false,
1881
+ "open_weights": true,
1584
1882
  "limit": {
1585
1883
  "context": 128000,
1586
- "output": 4096
1884
+ "output": 8192
1587
1885
  },
1588
1886
  "cost": {
1589
1887
  "input": 0.2,
@@ -1601,7 +1899,7 @@
1601
1899
  "tool_call": true,
1602
1900
  "structured_output": true,
1603
1901
  "temperature": true,
1604
- "release_date": "2025-12-23",
1902
+ "release_date": "2025-12-15",
1605
1903
  "last_updated": "2025-12-23",
1606
1904
  "modalities": {
1607
1905
  "input": [
@@ -1613,8 +1911,8 @@
1613
1911
  },
1614
1912
  "open_weights": true,
1615
1913
  "limit": {
1616
- "context": 128000,
1617
- "output": 4096
1914
+ "context": 262144,
1915
+ "output": 8192
1618
1916
  },
1619
1917
  "cost": {
1620
1918
  "input": 0.06,
@@ -1629,6 +1927,9 @@
1629
1927
  "attachment": true,
1630
1928
  "reasoning": true,
1631
1929
  "reasoning_options": [
1930
+ {
1931
+ "type": "toggle"
1932
+ },
1632
1933
  {
1633
1934
  "type": "effort",
1634
1935
  "values": [
@@ -1645,9 +1946,9 @@
1645
1946
  "tool_call": true,
1646
1947
  "structured_output": true,
1647
1948
  "temperature": true,
1648
- "knowledge": "2025-03-31",
1649
- "release_date": "2025-11-24",
1650
- "last_updated": "2025-08-01",
1949
+ "knowledge": "2025-05",
1950
+ "release_date": "2025-11-01",
1951
+ "last_updated": "2025-11-01",
1651
1952
  "modalities": {
1652
1953
  "input": [
1653
1954
  "text",
@@ -1672,14 +1973,15 @@
1672
1973
  },
1673
1974
  "minimax.minimax-m2.1": {
1674
1975
  "id": "minimax.minimax-m2.1",
1675
- "name": "MiniMax M2.1",
1676
- "description": "MiniMax model for chat, coding, office work, and agentic tasks",
1976
+ "name": "MiniMax-M2.1",
1977
+ "description": "Earlier MiniMax agent model for practical coding and productivity tasks",
1677
1978
  "family": "minimax",
1678
1979
  "attachment": false,
1679
1980
  "reasoning": true,
1680
1981
  "reasoning_options": [],
1681
1982
  "tool_call": true,
1682
- "structured_output": false,
1983
+ "interleaved": true,
1984
+ "structured_output": true,
1683
1985
  "temperature": true,
1684
1986
  "release_date": "2025-12-23",
1685
1987
  "last_updated": "2025-12-23",
@@ -1701,10 +2003,76 @@
1701
2003
  "output": 1.2
1702
2004
  }
1703
2005
  },
2006
+ "us.openai.gpt-6-luna": {
2007
+ "id": "us.openai.gpt-6-luna",
2008
+ "name": "GPT-6 Luna (US)",
2009
+ "description": "OpenAI's most efficient model for focused, high-volume tasks",
2010
+ "family": "gpt-luna",
2011
+ "attachment": true,
2012
+ "reasoning": true,
2013
+ "reasoning_options": [
2014
+ {
2015
+ "type": "effort",
2016
+ "values": [
2017
+ "none",
2018
+ "low",
2019
+ "medium",
2020
+ "high",
2021
+ "xhigh",
2022
+ "max"
2023
+ ]
2024
+ }
2025
+ ],
2026
+ "tool_call": true,
2027
+ "structured_output": true,
2028
+ "temperature": false,
2029
+ "knowledge": "2026-05-18",
2030
+ "release_date": "2026-09-22",
2031
+ "last_updated": "2026-09-22",
2032
+ "modalities": {
2033
+ "input": [
2034
+ "text",
2035
+ "image"
2036
+ ],
2037
+ "output": [
2038
+ "text"
2039
+ ]
2040
+ },
2041
+ "open_weights": false,
2042
+ "limit": {
2043
+ "context": 1050000,
2044
+ "input": 922000,
2045
+ "output": 128000
2046
+ },
2047
+ "cost": {
2048
+ "input": 0.11,
2049
+ "output": 0.55,
2050
+ "cache_read": 0.011,
2051
+ "cache_write": 0.1375,
2052
+ "tiers": [
2053
+ {
2054
+ "input": 0.22,
2055
+ "output": 0.825,
2056
+ "cache_read": 0.022,
2057
+ "cache_write": 0.275,
2058
+ "tier": {
2059
+ "type": "context",
2060
+ "size": 272000
2061
+ }
2062
+ }
2063
+ ],
2064
+ "context_over_200k": {
2065
+ "input": 0.22,
2066
+ "output": 0.825,
2067
+ "cache_read": 0.022,
2068
+ "cache_write": 0.275
2069
+ }
2070
+ }
2071
+ },
1704
2072
  "meta.llama3-3-70b-instruct-v1:0": {
1705
2073
  "id": "meta.llama3-3-70b-instruct-v1:0",
1706
2074
  "name": "Llama 3.3 70B Instruct",
1707
- "description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
2075
+ "description": "Popular open Llama workhorse for multilingual chat, coding, and self-hosting",
1708
2076
  "family": "llama",
1709
2077
  "attachment": false,
1710
2078
  "reasoning": false,
@@ -1734,16 +2102,20 @@
1734
2102
  "deepseek.v3-v1:0": {
1735
2103
  "id": "deepseek.v3-v1:0",
1736
2104
  "name": "DeepSeek-V3.1",
1737
- "description": "DeepSeek chat model for instruction following, coding, and analysis",
2105
+ "description": "Hybrid-reasoning DeepSeek model with thinking and non-thinking modes",
1738
2106
  "family": "deepseek",
1739
2107
  "attachment": false,
1740
2108
  "reasoning": true,
1741
- "reasoning_options": [],
2109
+ "reasoning_options": [
2110
+ {
2111
+ "type": "toggle"
2112
+ }
2113
+ ],
1742
2114
  "tool_call": true,
2115
+ "interleaved": true,
1743
2116
  "structured_output": true,
1744
2117
  "temperature": true,
1745
- "knowledge": "2024-07",
1746
- "release_date": "2025-09-18",
2118
+ "release_date": "2025-08-21",
1747
2119
  "last_updated": "2025-09-18",
1748
2120
  "modalities": {
1749
2121
  "input": [
@@ -1771,6 +2143,9 @@
1771
2143
  "attachment": true,
1772
2144
  "reasoning": true,
1773
2145
  "reasoning_options": [
2146
+ {
2147
+ "type": "toggle"
2148
+ },
1774
2149
  {
1775
2150
  "type": "effort",
1776
2151
  "values": [
@@ -1832,7 +2207,7 @@
1832
2207
  }
1833
2208
  ],
1834
2209
  "tool_call": true,
1835
- "structured_output": true,
2210
+ "structured_output": false,
1836
2211
  "temperature": false,
1837
2212
  "knowledge": "2026-01-31",
1838
2213
  "release_date": "2026-06-30",
@@ -1882,6 +2257,7 @@
1882
2257
  "open_weights": false,
1883
2258
  "limit": {
1884
2259
  "context": 1040000,
2260
+ "input": 1040000,
1885
2261
  "output": 8192
1886
2262
  },
1887
2263
  "cost": {
@@ -1889,6 +2265,53 @@
1889
2265
  "output": 6
1890
2266
  }
1891
2267
  },
2268
+ "google.gemma-4-e2b": {
2269
+ "id": "google.gemma-4-e2b",
2270
+ "name": "Gemma 4 E2B IT",
2271
+ "description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
2272
+ "family": "gemma",
2273
+ "attachment": true,
2274
+ "reasoning": true,
2275
+ "reasoning_options": [
2276
+ {
2277
+ "type": "effort",
2278
+ "values": [
2279
+ "none",
2280
+ "high"
2281
+ ]
2282
+ }
2283
+ ],
2284
+ "tool_call": true,
2285
+ "structured_output": true,
2286
+ "temperature": true,
2287
+ "release_date": "2026-04-02",
2288
+ "last_updated": "2026-04-02",
2289
+ "modalities": {
2290
+ "input": [
2291
+ "text",
2292
+ "image",
2293
+ "audio",
2294
+ "video"
2295
+ ],
2296
+ "output": [
2297
+ "text"
2298
+ ]
2299
+ },
2300
+ "open_weights": true,
2301
+ "limit": {
2302
+ "context": 131072,
2303
+ "output": 8192
2304
+ },
2305
+ "provider": {
2306
+ "npm": "@ai-sdk/amazon-bedrock/mantle",
2307
+ "api": "https://bedrock-mantle.${AWS_REGION}.api.aws/openai/v1",
2308
+ "shape": "responses"
2309
+ },
2310
+ "cost": {
2311
+ "input": 0.04,
2312
+ "output": 0.08
2313
+ }
2314
+ },
1892
2315
  "us.meta.llama4-maverick-17b-instruct-v1:0": {
1893
2316
  "id": "us.meta.llama4-maverick-17b-instruct-v1:0",
1894
2317
  "name": "Llama 4 Maverick 17B Instruct (US)",
@@ -1913,7 +2336,7 @@
1913
2336
  "open_weights": true,
1914
2337
  "limit": {
1915
2338
  "context": 1000000,
1916
- "output": 16384
2339
+ "output": 8192
1917
2340
  },
1918
2341
  "cost": {
1919
2342
  "input": 0.24,
@@ -1923,7 +2346,7 @@
1923
2346
  "meta.llama3-1-8b-instruct-v1:0": {
1924
2347
  "id": "meta.llama3-1-8b-instruct-v1:0",
1925
2348
  "name": "Llama 3.1 8B Instruct",
1926
- "description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
2349
+ "description": "Compact open Llama model for lightweight chat, drafting, and self-hosting",
1927
2350
  "family": "llama",
1928
2351
  "attachment": false,
1929
2352
  "reasoning": false,
@@ -1952,14 +2375,15 @@
1952
2375
  },
1953
2376
  "minimax.minimax-m2": {
1954
2377
  "id": "minimax.minimax-m2",
1955
- "name": "MiniMax M2",
1956
- "description": "MiniMax model for chat, coding, office work, and agentic tasks",
2378
+ "name": "MiniMax-M2",
2379
+ "description": "Efficient open MiniMax model built for coding agents and tool-heavy workflows",
1957
2380
  "family": "minimax",
1958
2381
  "attachment": false,
1959
2382
  "reasoning": true,
1960
2383
  "reasoning_options": [],
1961
2384
  "tool_call": true,
1962
- "structured_output": false,
2385
+ "interleaved": true,
2386
+ "structured_output": true,
1963
2387
  "temperature": true,
1964
2388
  "release_date": "2025-10-27",
1965
2389
  "last_updated": "2025-10-27",
@@ -1981,6 +2405,52 @@
1981
2405
  "output": 1.2
1982
2406
  }
1983
2407
  },
2408
+ "au.anthropic.claude-opus-5-5": {
2409
+ "id": "au.anthropic.claude-opus-5-5",
2410
+ "name": "Claude Opus 5.5 (AU)",
2411
+ "description": "Claude model for long-running agentic coding and knowledge work",
2412
+ "family": "claude-opus",
2413
+ "attachment": true,
2414
+ "reasoning": true,
2415
+ "reasoning_options": [
2416
+ {
2417
+ "type": "effort",
2418
+ "values": [
2419
+ "low",
2420
+ "medium",
2421
+ "high",
2422
+ "xhigh",
2423
+ "max"
2424
+ ]
2425
+ }
2426
+ ],
2427
+ "tool_call": true,
2428
+ "temperature": false,
2429
+ "knowledge": "2026-06",
2430
+ "release_date": "2026-09-22",
2431
+ "last_updated": "2026-09-22",
2432
+ "modalities": {
2433
+ "input": [
2434
+ "text",
2435
+ "image",
2436
+ "pdf"
2437
+ ],
2438
+ "output": [
2439
+ "text"
2440
+ ]
2441
+ },
2442
+ "open_weights": false,
2443
+ "limit": {
2444
+ "context": 1000000,
2445
+ "output": 128000
2446
+ },
2447
+ "cost": {
2448
+ "input": 4.4,
2449
+ "output": 22,
2450
+ "cache_read": 0.22,
2451
+ "cache_write": 5.5
2452
+ }
2453
+ },
1984
2454
  "global.anthropic.claude-opus-5": {
1985
2455
  "id": "global.anthropic.claude-opus-5",
1986
2456
  "name": "Claude Opus 5 (Global)",
@@ -1989,6 +2459,9 @@
1989
2459
  "attachment": true,
1990
2460
  "reasoning": true,
1991
2461
  "reasoning_options": [
2462
+ {
2463
+ "type": "toggle"
2464
+ },
1992
2465
  {
1993
2466
  "type": "effort",
1994
2467
  "values": [
@@ -2027,19 +2500,68 @@
2027
2500
  "cache_write": 6.25
2028
2501
  }
2029
2502
  },
2503
+ "eu.anthropic.claude-sonnet-4-20250514-v1:0": {
2504
+ "id": "eu.anthropic.claude-sonnet-4-20250514-v1:0",
2505
+ "name": "Claude Sonnet 4 (EU)",
2506
+ "description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
2507
+ "family": "claude-sonnet",
2508
+ "attachment": true,
2509
+ "reasoning": true,
2510
+ "reasoning_options": [
2511
+ {
2512
+ "type": "toggle"
2513
+ },
2514
+ {
2515
+ "type": "budget_tokens",
2516
+ "min": 1024
2517
+ }
2518
+ ],
2519
+ "tool_call": true,
2520
+ "temperature": true,
2521
+ "knowledge": "2025-03-31",
2522
+ "release_date": "2025-05-22",
2523
+ "last_updated": "2025-05-22",
2524
+ "modalities": {
2525
+ "input": [
2526
+ "text",
2527
+ "image",
2528
+ "pdf"
2529
+ ],
2530
+ "output": [
2531
+ "text"
2532
+ ]
2533
+ },
2534
+ "open_weights": false,
2535
+ "limit": {
2536
+ "context": 200000,
2537
+ "output": 64000
2538
+ },
2539
+ "status": "deprecated",
2540
+ "cost": {
2541
+ "input": 3,
2542
+ "output": 15,
2543
+ "cache_read": 0.3,
2544
+ "cache_write": 3.75
2545
+ }
2546
+ },
2030
2547
  "qwen.qwen3-32b-v1:0": {
2031
2548
  "id": "qwen.qwen3-32b-v1:0",
2032
- "name": "Qwen3 32B (dense)",
2033
- "description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
2549
+ "name": "Qwen3 32B",
2550
+ "description": "Dense open Qwen model for self-hosted chat, reasoning, and coding",
2034
2551
  "family": "qwen",
2035
2552
  "attachment": false,
2036
2553
  "reasoning": true,
2037
- "reasoning_options": [],
2554
+ "reasoning_options": [
2555
+ {
2556
+ "type": "toggle"
2557
+ }
2558
+ ],
2038
2559
  "tool_call": true,
2560
+ "interleaved": true,
2039
2561
  "structured_output": true,
2040
2562
  "temperature": true,
2041
- "knowledge": "2024-04",
2042
- "release_date": "2025-09-18",
2563
+ "knowledge": "2025-04",
2564
+ "release_date": "2025-04",
2043
2565
  "last_updated": "2025-09-18",
2044
2566
  "modalities": {
2045
2567
  "input": [
@@ -2051,7 +2573,7 @@
2051
2573
  },
2052
2574
  "open_weights": true,
2053
2575
  "limit": {
2054
- "context": 16384,
2576
+ "context": 32768,
2055
2577
  "output": 16384
2056
2578
  },
2057
2579
  "cost": {
@@ -2062,14 +2584,14 @@
2062
2584
  "writer.palmyra-x4-v1:0": {
2063
2585
  "id": "writer.palmyra-x4-v1:0",
2064
2586
  "name": "Palmyra X4",
2065
- "description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use",
2587
+ "description": "Enterprise language model for workflow automation, coding, data analysis, and tool use",
2066
2588
  "family": "palmyra",
2067
2589
  "attachment": false,
2068
2590
  "reasoning": true,
2069
2591
  "reasoning_options": [],
2070
2592
  "tool_call": true,
2071
2593
  "temperature": true,
2072
- "release_date": "2025-04-28",
2594
+ "release_date": "2024-10-09",
2073
2595
  "last_updated": "2025-04-28",
2074
2596
  "modalities": {
2075
2597
  "input": [
@@ -2105,7 +2627,8 @@
2105
2627
  "input": [
2106
2628
  "text",
2107
2629
  "image",
2108
- "video"
2630
+ "video",
2631
+ "pdf"
2109
2632
  ],
2110
2633
  "output": [
2111
2634
  "text"
@@ -2114,27 +2637,28 @@
2114
2637
  "open_weights": false,
2115
2638
  "limit": {
2116
2639
  "context": 300000,
2117
- "output": 8192
2640
+ "output": 10000
2118
2641
  },
2119
2642
  "cost": {
2120
2643
  "input": 0.8,
2121
2644
  "output": 3.2,
2122
- "cache_read": 0.2
2645
+ "cache_read": 0.2,
2646
+ "cache_write": 0.8
2123
2647
  }
2124
2648
  },
2125
2649
  "google.gemma-3-12b-it": {
2126
2650
  "id": "google.gemma-3-12b-it",
2127
- "name": "Google Gemma 3 12B",
2128
- "description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
2651
+ "name": "Gemma 3 12B IT",
2652
+ "description": "Open multimodal Gemma instruction model for multilingual text generation and image understanding",
2129
2653
  "family": "gemma",
2130
- "attachment": false,
2654
+ "attachment": true,
2131
2655
  "reasoning": false,
2132
2656
  "tool_call": false,
2133
2657
  "structured_output": true,
2134
2658
  "temperature": true,
2135
- "knowledge": "2024-12",
2136
- "release_date": "2024-12-01",
2137
- "last_updated": "2024-12-01",
2659
+ "knowledge": "2024-08",
2660
+ "release_date": "2025-03-12",
2661
+ "last_updated": "2025-03-12",
2138
2662
  "modalities": {
2139
2663
  "input": [
2140
2664
  "text",
@@ -2144,14 +2668,14 @@
2144
2668
  "text"
2145
2669
  ]
2146
2670
  },
2147
- "open_weights": false,
2671
+ "open_weights": true,
2148
2672
  "limit": {
2149
2673
  "context": 131072,
2150
2674
  "output": 8192
2151
2675
  },
2152
2676
  "cost": {
2153
- "input": 0.049999999999999996,
2154
- "output": 0.09999999999999999
2677
+ "input": 0.09,
2678
+ "output": 0.29
2155
2679
  }
2156
2680
  },
2157
2681
  "au.anthropic.claude-opus-4-8": {
@@ -2162,6 +2686,9 @@
2162
2686
  "attachment": true,
2163
2687
  "reasoning": true,
2164
2688
  "reasoning_options": [
2689
+ {
2690
+ "type": "toggle"
2691
+ },
2165
2692
  {
2166
2693
  "type": "effort",
2167
2694
  "values": [
@@ -2194,10 +2721,10 @@
2194
2721
  "output": 128000
2195
2722
  },
2196
2723
  "cost": {
2197
- "input": 5,
2198
- "output": 25,
2199
- "cache_read": 0.5,
2200
- "cache_write": 6.25
2724
+ "input": 5.5,
2725
+ "output": 27.5,
2726
+ "cache_read": 0.55,
2727
+ "cache_write": 6.875
2201
2728
  }
2202
2729
  },
2203
2730
  "jp.anthropic.claude-haiku-4-5-20251001-v1:0": {
@@ -2208,6 +2735,9 @@
2208
2735
  "attachment": true,
2209
2736
  "reasoning": true,
2210
2737
  "reasoning_options": [
2738
+ {
2739
+ "type": "toggle"
2740
+ },
2211
2741
  {
2212
2742
  "type": "budget_tokens",
2213
2743
  "min": 1024
@@ -2235,10 +2765,10 @@
2235
2765
  "output": 64000
2236
2766
  },
2237
2767
  "cost": {
2238
- "input": 1,
2239
- "output": 5,
2240
- "cache_read": 0.1,
2241
- "cache_write": 1.25
2768
+ "input": 1.1,
2769
+ "output": 5.5,
2770
+ "cache_read": 0.11,
2771
+ "cache_write": 1.375
2242
2772
  }
2243
2773
  },
2244
2774
  "eu.amazon.nova-2-lite-v1:0": {
@@ -2263,7 +2793,8 @@
2263
2793
  ],
2264
2794
  "tool_call": true,
2265
2795
  "temperature": true,
2266
- "release_date": "2025-12-01",
2796
+ "knowledge": "2025-10",
2797
+ "release_date": "2025-12-02",
2267
2798
  "last_updated": "2025-12-01",
2268
2799
  "modalities": {
2269
2800
  "input": [
@@ -2279,12 +2810,13 @@
2279
2810
  "open_weights": false,
2280
2811
  "limit": {
2281
2812
  "context": 1000000,
2282
- "output": 64000
2813
+ "output": 65535
2283
2814
  },
2284
2815
  "cost": {
2285
2816
  "input": 0.374,
2286
2817
  "output": 3.157,
2287
- "cache_read": 0.0935
2818
+ "cache_read": 0.0935,
2819
+ "cache_write": 0.374
2288
2820
  }
2289
2821
  },
2290
2822
  "eu.anthropic.claude-opus-4-8": {
@@ -2295,6 +2827,9 @@
2295
2827
  "attachment": true,
2296
2828
  "reasoning": true,
2297
2829
  "reasoning_options": [
2830
+ {
2831
+ "type": "toggle"
2832
+ },
2298
2833
  {
2299
2834
  "type": "effort",
2300
2835
  "values": [
@@ -2341,6 +2876,9 @@
2341
2876
  "attachment": true,
2342
2877
  "reasoning": true,
2343
2878
  "reasoning_options": [
2879
+ {
2880
+ "type": "toggle"
2881
+ },
2344
2882
  {
2345
2883
  "type": "effort",
2346
2884
  "values": [
@@ -2373,33 +2911,34 @@
2373
2911
  "output": 128000
2374
2912
  },
2375
2913
  "cost": {
2376
- "input": 5,
2377
- "output": 25,
2378
- "cache_read": 0.5,
2379
- "cache_write": 6.25
2914
+ "input": 5.5,
2915
+ "output": 27.5,
2916
+ "cache_read": 0.55,
2917
+ "cache_write": 6.875
2380
2918
  }
2381
2919
  },
2382
2920
  "mistral.ministral-3-14b-instruct": {
2383
2921
  "id": "mistral.ministral-3-14b-instruct",
2384
2922
  "name": "Ministral 14B 3.0",
2385
- "description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
2923
+ "description": "Open vision-language model for efficient local deployment, instruction following, and tool use",
2386
2924
  "family": "ministral",
2387
- "attachment": false,
2925
+ "attachment": true,
2388
2926
  "reasoning": false,
2389
2927
  "tool_call": true,
2390
2928
  "structured_output": true,
2391
2929
  "temperature": true,
2392
- "release_date": "2024-12-01",
2393
- "last_updated": "2024-12-01",
2930
+ "release_date": "2025-12-02",
2931
+ "last_updated": "2025-12-02",
2394
2932
  "modalities": {
2395
2933
  "input": [
2396
- "text"
2934
+ "text",
2935
+ "image"
2397
2936
  ],
2398
2937
  "output": [
2399
2938
  "text"
2400
2939
  ]
2401
2940
  },
2402
- "open_weights": false,
2941
+ "open_weights": true,
2403
2942
  "limit": {
2404
2943
  "context": 128000,
2405
2944
  "output": 4096
@@ -2415,7 +2954,17 @@
2415
2954
  "description": "Safety model for policy screening, moderation, and risk-aware routing workflows",
2416
2955
  "family": "gpt-oss",
2417
2956
  "attachment": false,
2418
- "reasoning": false,
2957
+ "reasoning": true,
2958
+ "reasoning_options": [
2959
+ {
2960
+ "type": "effort",
2961
+ "values": [
2962
+ "low",
2963
+ "medium",
2964
+ "high"
2965
+ ]
2966
+ }
2967
+ ],
2419
2968
  "tool_call": true,
2420
2969
  "structured_output": true,
2421
2970
  "temperature": true,
@@ -2439,6 +2988,50 @@
2439
2988
  "output": 0.2
2440
2989
  }
2441
2990
  },
2991
+ "global.anthropic.claude-sonnet-4-20250514-v1:0": {
2992
+ "id": "global.anthropic.claude-sonnet-4-20250514-v1:0",
2993
+ "name": "Claude Sonnet 4 (Global)",
2994
+ "description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
2995
+ "family": "claude-sonnet",
2996
+ "attachment": true,
2997
+ "reasoning": true,
2998
+ "reasoning_options": [
2999
+ {
3000
+ "type": "toggle"
3001
+ },
3002
+ {
3003
+ "type": "budget_tokens",
3004
+ "min": 1024
3005
+ }
3006
+ ],
3007
+ "tool_call": true,
3008
+ "temperature": true,
3009
+ "knowledge": "2025-03-31",
3010
+ "release_date": "2025-05-22",
3011
+ "last_updated": "2025-05-22",
3012
+ "modalities": {
3013
+ "input": [
3014
+ "text",
3015
+ "image",
3016
+ "pdf"
3017
+ ],
3018
+ "output": [
3019
+ "text"
3020
+ ]
3021
+ },
3022
+ "open_weights": false,
3023
+ "limit": {
3024
+ "context": 200000,
3025
+ "output": 64000
3026
+ },
3027
+ "status": "deprecated",
3028
+ "cost": {
3029
+ "input": 3,
3030
+ "output": 15,
3031
+ "cache_read": 0.3,
3032
+ "cache_write": 3.75
3033
+ }
3034
+ },
2442
3035
  "global.amazon.nova-2-lite-v1:0": {
2443
3036
  "id": "global.amazon.nova-2-lite-v1:0",
2444
3037
  "name": "Nova 2 Lite (Global)",
@@ -2461,7 +3054,8 @@
2461
3054
  ],
2462
3055
  "tool_call": true,
2463
3056
  "temperature": true,
2464
- "release_date": "2025-12-01",
3057
+ "knowledge": "2025-10",
3058
+ "release_date": "2025-12-02",
2465
3059
  "last_updated": "2025-12-01",
2466
3060
  "modalities": {
2467
3061
  "input": [
@@ -2477,12 +3071,13 @@
2477
3071
  "open_weights": false,
2478
3072
  "limit": {
2479
3073
  "context": 1000000,
2480
- "output": 64000
3074
+ "output": 65535
2481
3075
  },
2482
3076
  "cost": {
2483
3077
  "input": 0.3,
2484
3078
  "output": 2.5,
2485
- "cache_read": 0.075
3079
+ "cache_read": 0.075,
3080
+ "cache_write": 0.3
2486
3081
  }
2487
3082
  },
2488
3083
  "eu.amazon.nova-micro-v1:0": {
@@ -2508,12 +3103,13 @@
2508
3103
  "open_weights": false,
2509
3104
  "limit": {
2510
3105
  "context": 128000,
2511
- "output": 8192
3106
+ "output": 10000
2512
3107
  },
2513
3108
  "cost": {
2514
3109
  "input": 0.04,
2515
3110
  "output": 0.16,
2516
- "cache_read": 0.01
3111
+ "cache_read": 0.01,
3112
+ "cache_write": 0.04
2517
3113
  }
2518
3114
  },
2519
3115
  "openai.gpt-5.6-luna": {
@@ -2596,6 +3192,9 @@
2596
3192
  "attachment": true,
2597
3193
  "reasoning": true,
2598
3194
  "reasoning_options": [
3195
+ {
3196
+ "type": "toggle"
3197
+ },
2599
3198
  {
2600
3199
  "type": "effort",
2601
3200
  "values": [
@@ -2632,16 +3231,16 @@
2632
3231
  "output": 128000
2633
3232
  },
2634
3233
  "cost": {
2635
- "input": 5,
2636
- "output": 25,
2637
- "cache_read": 0.5,
2638
- "cache_write": 6.25
3234
+ "input": 5.5,
3235
+ "output": 27.5,
3236
+ "cache_read": 0.55,
3237
+ "cache_write": 6.875
2639
3238
  }
2640
3239
  },
2641
3240
  "openai.gpt-oss-20b-1:0": {
2642
3241
  "id": "openai.gpt-oss-20b-1:0",
2643
3242
  "name": "gpt-oss-20b",
2644
- "description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
3243
+ "description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
2645
3244
  "family": "gpt-oss",
2646
3245
  "attachment": false,
2647
3246
  "reasoning": true,
@@ -2656,39 +3255,77 @@
2656
3255
  }
2657
3256
  ],
2658
3257
  "tool_call": true,
2659
- "structured_output": true,
3258
+ "structured_output": true,
3259
+ "temperature": true,
3260
+ "release_date": "2025-08-05",
3261
+ "last_updated": "2025-08-05",
3262
+ "modalities": {
3263
+ "input": [
3264
+ "text"
3265
+ ],
3266
+ "output": [
3267
+ "text"
3268
+ ]
3269
+ },
3270
+ "open_weights": true,
3271
+ "limit": {
3272
+ "context": 128000,
3273
+ "output": 16384
3274
+ },
3275
+ "cost": {
3276
+ "input": 0.07,
3277
+ "output": 0.3
3278
+ }
3279
+ },
3280
+ "us.amazon.nova-premier-v1:0": {
3281
+ "id": "us.amazon.nova-premier-v1:0",
3282
+ "name": "Nova Premier (US)",
3283
+ "description": "Multimodal model for complex analysis, long-context understanding, tool use, and model distillation",
3284
+ "family": "nova",
3285
+ "attachment": true,
3286
+ "reasoning": false,
3287
+ "tool_call": true,
3288
+ "structured_output": false,
2660
3289
  "temperature": true,
2661
- "release_date": "2025-08-05",
2662
- "last_updated": "2025-08-05",
3290
+ "knowledge": "2024-10",
3291
+ "release_date": "2025-04-30",
3292
+ "last_updated": "2025-04-30",
2663
3293
  "modalities": {
2664
3294
  "input": [
2665
- "text"
3295
+ "text",
3296
+ "image",
3297
+ "video",
3298
+ "pdf"
2666
3299
  ],
2667
3300
  "output": [
2668
3301
  "text"
2669
3302
  ]
2670
3303
  },
2671
- "open_weights": true,
3304
+ "open_weights": false,
2672
3305
  "limit": {
2673
- "context": 128000,
2674
- "output": 16384
3306
+ "context": 1000000,
3307
+ "output": 10000
2675
3308
  },
3309
+ "status": "deprecated",
2676
3310
  "cost": {
2677
- "input": 0.07,
2678
- "output": 0.3
3311
+ "input": 2.5,
3312
+ "output": 12.5,
3313
+ "cache_read": 0.625,
3314
+ "cache_write": 2.5
2679
3315
  }
2680
3316
  },
2681
3317
  "qwen.qwen3-vl-235b-a22b": {
2682
3318
  "id": "qwen.qwen3-vl-235b-a22b",
2683
- "name": "Qwen/Qwen3-VL-235B-A22B-Instruct",
2684
- "description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
3319
+ "name": "Qwen3 VL 235B A22B Instruct",
3320
+ "description": "Qwen vision-language instruct model for visual reasoning, documents, and agent tasks",
2685
3321
  "family": "qwen",
2686
3322
  "attachment": true,
2687
3323
  "reasoning": false,
2688
3324
  "tool_call": true,
2689
3325
  "structured_output": true,
2690
3326
  "temperature": true,
2691
- "release_date": "2025-10-04",
3327
+ "knowledge": "2025-03-31",
3328
+ "release_date": "2025-09-23",
2692
3329
  "last_updated": "2025-11-25",
2693
3330
  "modalities": {
2694
3331
  "input": [
@@ -2699,14 +3336,14 @@
2699
3336
  "text"
2700
3337
  ]
2701
3338
  },
2702
- "open_weights": false,
3339
+ "open_weights": true,
2703
3340
  "limit": {
2704
- "context": 262000,
3341
+ "context": 262144,
2705
3342
  "output": 262000
2706
3343
  },
2707
3344
  "cost": {
2708
- "input": 0.3,
2709
- "output": 1.5
3345
+ "input": 0.53,
3346
+ "output": 2.66
2710
3347
  }
2711
3348
  },
2712
3349
  "amazon.nova-2-lite-v1:0": {
@@ -2714,7 +3351,7 @@
2714
3351
  "name": "Nova 2 Lite",
2715
3352
  "description": "Multimodal reasoning model for visual analysis, planning, and tool use",
2716
3353
  "family": "nova",
2717
- "attachment": false,
3354
+ "attachment": true,
2718
3355
  "reasoning": true,
2719
3356
  "reasoning_options": [
2720
3357
  {
@@ -2731,13 +3368,15 @@
2731
3368
  ],
2732
3369
  "tool_call": true,
2733
3370
  "temperature": true,
2734
- "release_date": "2024-12-01",
2735
- "last_updated": "2024-12-01",
3371
+ "knowledge": "2025-10",
3372
+ "release_date": "2025-12-02",
3373
+ "last_updated": "2025-12-01",
2736
3374
  "modalities": {
2737
3375
  "input": [
2738
3376
  "text",
2739
3377
  "image",
2740
- "video"
3378
+ "video",
3379
+ "pdf"
2741
3380
  ],
2742
3381
  "output": [
2743
3382
  "text"
@@ -2745,12 +3384,80 @@
2745
3384
  },
2746
3385
  "open_weights": false,
2747
3386
  "limit": {
2748
- "context": 128000,
2749
- "output": 4096
3387
+ "context": 1000000,
3388
+ "output": 65535
2750
3389
  },
2751
3390
  "cost": {
2752
3391
  "input": 0.33,
2753
- "output": 2.75
3392
+ "output": 2.75,
3393
+ "cache_read": 0.0825,
3394
+ "cache_write": 0.33
3395
+ }
3396
+ },
3397
+ "global.openai.gpt-6-sol": {
3398
+ "id": "global.openai.gpt-6-sol",
3399
+ "name": "GPT-6 Sol (Global)",
3400
+ "description": "OpenAI model for complex coding and agentic workflows",
3401
+ "family": "gpt-sol",
3402
+ "attachment": true,
3403
+ "reasoning": true,
3404
+ "reasoning_options": [
3405
+ {
3406
+ "type": "effort",
3407
+ "values": [
3408
+ "none",
3409
+ "low",
3410
+ "medium",
3411
+ "high",
3412
+ "xhigh",
3413
+ "max"
3414
+ ]
3415
+ }
3416
+ ],
3417
+ "tool_call": true,
3418
+ "structured_output": true,
3419
+ "temperature": false,
3420
+ "knowledge": "2026-04-20",
3421
+ "release_date": "2026-09-22",
3422
+ "last_updated": "2026-09-22",
3423
+ "modalities": {
3424
+ "input": [
3425
+ "text",
3426
+ "image"
3427
+ ],
3428
+ "output": [
3429
+ "text"
3430
+ ]
3431
+ },
3432
+ "open_weights": false,
3433
+ "limit": {
3434
+ "context": 1050000,
3435
+ "input": 922000,
3436
+ "output": 128000
3437
+ },
3438
+ "cost": {
3439
+ "input": 2,
3440
+ "output": 10,
3441
+ "cache_read": 0.2,
3442
+ "cache_write": 2.5,
3443
+ "tiers": [
3444
+ {
3445
+ "input": 4,
3446
+ "output": 15,
3447
+ "cache_read": 0.4,
3448
+ "cache_write": 5,
3449
+ "tier": {
3450
+ "type": "context",
3451
+ "size": 272000
3452
+ }
3453
+ }
3454
+ ],
3455
+ "context_over_200k": {
3456
+ "input": 4,
3457
+ "output": 15,
3458
+ "cache_read": 0.4,
3459
+ "cache_write": 5
3460
+ }
2754
3461
  }
2755
3462
  },
2756
3463
  "global.xai.grok-4.6": {
@@ -2805,6 +3512,9 @@
2805
3512
  "attachment": true,
2806
3513
  "reasoning": true,
2807
3514
  "reasoning_options": [
3515
+ {
3516
+ "type": "toggle"
3517
+ },
2808
3518
  {
2809
3519
  "type": "effort",
2810
3520
  "values": [
@@ -2821,9 +3531,9 @@
2821
3531
  "tool_call": true,
2822
3532
  "structured_output": true,
2823
3533
  "temperature": true,
2824
- "knowledge": "2025-03-31",
2825
- "release_date": "2025-11-24",
2826
- "last_updated": "2025-08-01",
3534
+ "knowledge": "2025-05",
3535
+ "release_date": "2025-11-01",
3536
+ "last_updated": "2025-11-01",
2827
3537
  "modalities": {
2828
3538
  "input": [
2829
3539
  "text",
@@ -2862,7 +3572,8 @@
2862
3572
  "input": [
2863
3573
  "text",
2864
3574
  "image",
2865
- "video"
3575
+ "video",
3576
+ "pdf"
2866
3577
  ],
2867
3578
  "output": [
2868
3579
  "text"
@@ -2871,12 +3582,13 @@
2871
3582
  "open_weights": false,
2872
3583
  "limit": {
2873
3584
  "context": 300000,
2874
- "output": 8192
3585
+ "output": 10000
2875
3586
  },
2876
3587
  "cost": {
2877
3588
  "input": 0.06,
2878
3589
  "output": 0.24,
2879
- "cache_read": 0.015
3590
+ "cache_read": 0.015,
3591
+ "cache_write": 0.06
2880
3592
  }
2881
3593
  },
2882
3594
  "anthropic.claude-opus-4-8": {
@@ -2887,6 +3599,9 @@
2887
3599
  "attachment": true,
2888
3600
  "reasoning": true,
2889
3601
  "reasoning_options": [
3602
+ {
3603
+ "type": "toggle"
3604
+ },
2890
3605
  {
2891
3606
  "type": "effort",
2892
3607
  "values": [
@@ -2947,7 +3662,8 @@
2947
3662
  ],
2948
3663
  "tool_call": true,
2949
3664
  "temperature": true,
2950
- "release_date": "2025-12-01",
3665
+ "knowledge": "2025-10",
3666
+ "release_date": "2025-12-02",
2951
3667
  "last_updated": "2025-12-01",
2952
3668
  "modalities": {
2953
3669
  "input": [
@@ -2963,12 +3679,13 @@
2963
3679
  "open_weights": false,
2964
3680
  "limit": {
2965
3681
  "context": 1000000,
2966
- "output": 64000
3682
+ "output": 65535
2967
3683
  },
2968
3684
  "cost": {
2969
3685
  "input": 0.33,
2970
3686
  "output": 2.75,
2971
- "cache_read": 0.0825
3687
+ "cache_read": 0.0825,
3688
+ "cache_write": 0.33
2972
3689
  }
2973
3690
  },
2974
3691
  "us.openai.gpt-5.6-terra": {
@@ -3000,8 +3717,7 @@
3000
3717
  "modalities": {
3001
3718
  "input": [
3002
3719
  "text",
3003
- "image",
3004
- "pdf"
3720
+ "image"
3005
3721
  ],
3006
3722
  "output": [
3007
3723
  "text"
@@ -3114,7 +3830,8 @@
3114
3830
  "input": [
3115
3831
  "text",
3116
3832
  "image",
3117
- "video"
3833
+ "video",
3834
+ "pdf"
3118
3835
  ],
3119
3836
  "output": [
3120
3837
  "text"
@@ -3123,22 +3840,26 @@
3123
3840
  "open_weights": false,
3124
3841
  "limit": {
3125
3842
  "context": 300000,
3126
- "output": 8192
3843
+ "output": 10000
3127
3844
  },
3128
3845
  "cost": {
3129
3846
  "input": 0.8,
3130
3847
  "output": 3.2,
3131
- "cache_read": 0.2
3848
+ "cache_read": 0.2,
3849
+ "cache_write": 0.8
3132
3850
  }
3133
3851
  },
3134
3852
  "us.anthropic.claude-opus-4-7": {
3135
3853
  "id": "us.anthropic.claude-opus-4-7",
3136
3854
  "name": "Claude Opus 4.7 (US)",
3137
- "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
3855
+ "description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
3138
3856
  "family": "claude-opus",
3139
3857
  "attachment": true,
3140
3858
  "reasoning": true,
3141
3859
  "reasoning_options": [
3860
+ {
3861
+ "type": "toggle"
3862
+ },
3142
3863
  {
3143
3864
  "type": "effort",
3144
3865
  "values": [
@@ -3171,20 +3892,23 @@
3171
3892
  "output": 128000
3172
3893
  },
3173
3894
  "cost": {
3174
- "input": 5,
3175
- "output": 25,
3176
- "cache_read": 0.5,
3177
- "cache_write": 6.25
3895
+ "input": 5.5,
3896
+ "output": 27.5,
3897
+ "cache_read": 0.55,
3898
+ "cache_write": 6.875
3178
3899
  }
3179
3900
  },
3180
3901
  "au.anthropic.claude-opus-4-6-v1": {
3181
3902
  "id": "au.anthropic.claude-opus-4-6-v1",
3182
3903
  "name": "AU Anthropic Claude Opus 4.6",
3183
- "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
3904
+ "description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
3184
3905
  "family": "claude-opus",
3185
3906
  "attachment": true,
3186
3907
  "reasoning": true,
3187
3908
  "reasoning_options": [
3909
+ {
3910
+ "type": "toggle"
3911
+ },
3188
3912
  {
3189
3913
  "type": "effort",
3190
3914
  "values": [
@@ -3202,9 +3926,9 @@
3202
3926
  "tool_call": true,
3203
3927
  "structured_output": true,
3204
3928
  "temperature": true,
3205
- "knowledge": "2025-05",
3929
+ "knowledge": "2025-05-31",
3206
3930
  "release_date": "2026-02-05",
3207
- "last_updated": "2026-02-05",
3931
+ "last_updated": "2026-03-13",
3208
3932
  "modalities": {
3209
3933
  "input": [
3210
3934
  "text",
@@ -3221,10 +3945,10 @@
3221
3945
  "output": 128000
3222
3946
  },
3223
3947
  "cost": {
3224
- "input": 16.5,
3225
- "output": 82.5,
3226
- "cache_read": 1.65,
3227
- "cache_write": 20.625
3948
+ "input": 5.5,
3949
+ "output": 27.5,
3950
+ "cache_read": 0.55,
3951
+ "cache_write": 6.875
3228
3952
  }
3229
3953
  },
3230
3954
  "writer.palmyra-x5-v1:0": {
@@ -3250,6 +3974,7 @@
3250
3974
  "open_weights": false,
3251
3975
  "limit": {
3252
3976
  "context": 1040000,
3977
+ "input": 1040000,
3253
3978
  "output": 8192
3254
3979
  },
3255
3980
  "cost": {
@@ -3286,8 +4011,7 @@
3286
4011
  "modalities": {
3287
4012
  "input": [
3288
4013
  "text",
3289
- "image",
3290
- "pdf"
4014
+ "image"
3291
4015
  ],
3292
4016
  "output": [
3293
4017
  "text"
@@ -3404,6 +4128,9 @@
3404
4128
  "attachment": true,
3405
4129
  "reasoning": true,
3406
4130
  "reasoning_options": [
4131
+ {
4132
+ "type": "toggle"
4133
+ },
3407
4134
  {
3408
4135
  "type": "effort",
3409
4136
  "values": [
@@ -3444,16 +4171,17 @@
3444
4171
  },
3445
4172
  "minimax.minimax-m2.5": {
3446
4173
  "id": "minimax.minimax-m2.5",
3447
- "name": "MiniMax M2.5",
3448
- "description": "MiniMax model for chat, coding, office work, and agentic tasks",
4174
+ "name": "MiniMax-M2.5",
4175
+ "description": "Prior MiniMax coding model for agent workflows, office edits, and automation",
3449
4176
  "family": "minimax",
3450
4177
  "attachment": false,
3451
4178
  "reasoning": true,
3452
4179
  "reasoning_options": [],
3453
4180
  "tool_call": true,
3454
- "structured_output": false,
4181
+ "interleaved": true,
4182
+ "structured_output": true,
3455
4183
  "temperature": true,
3456
- "release_date": "2026-03-18",
4184
+ "release_date": "2026-02-12",
3457
4185
  "last_updated": "2026-03-18",
3458
4186
  "modalities": {
3459
4187
  "input": [
@@ -3476,7 +4204,7 @@
3476
4204
  "openai.gpt-oss-120b-1:0": {
3477
4205
  "id": "openai.gpt-oss-120b-1:0",
3478
4206
  "name": "gpt-oss-120b",
3479
- "description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
4207
+ "description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
3480
4208
  "family": "gpt-oss",
3481
4209
  "attachment": false,
3482
4210
  "reasoning": true,
@@ -3516,11 +4244,14 @@
3516
4244
  "eu.anthropic.claude-opus-4-7": {
3517
4245
  "id": "eu.anthropic.claude-opus-4-7",
3518
4246
  "name": "Claude Opus 4.7 (EU)",
3519
- "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
4247
+ "description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
3520
4248
  "family": "claude-opus",
3521
4249
  "attachment": true,
3522
4250
  "reasoning": true,
3523
4251
  "reasoning_options": [
4252
+ {
4253
+ "type": "toggle"
4254
+ },
3524
4255
  {
3525
4256
  "type": "effort",
3526
4257
  "values": [
@@ -3582,8 +4313,8 @@
3582
4313
  },
3583
4314
  "open_weights": true,
3584
4315
  "limit": {
3585
- "context": 3500000,
3586
- "output": 16384
4316
+ "context": 10000000,
4317
+ "output": 8192
3587
4318
  },
3588
4319
  "cost": {
3589
4320
  "input": 0.17,
@@ -3611,7 +4342,7 @@
3611
4342
  }
3612
4343
  ],
3613
4344
  "tool_call": true,
3614
- "structured_output": false,
4345
+ "structured_output": true,
3615
4346
  "temperature": false,
3616
4347
  "knowledge": "2026-02-16",
3617
4348
  "release_date": "2026-07-09",
@@ -3619,8 +4350,7 @@
3619
4350
  "modalities": {
3620
4351
  "input": [
3621
4352
  "text",
3622
- "image",
3623
- "pdf"
4353
+ "image"
3624
4354
  ],
3625
4355
  "output": [
3626
4356
  "text"
@@ -3657,10 +4387,54 @@
3657
4387
  }
3658
4388
  }
3659
4389
  },
4390
+ "us.anthropic.claude-sonnet-4-20250514-v1:0": {
4391
+ "id": "us.anthropic.claude-sonnet-4-20250514-v1:0",
4392
+ "name": "Claude Sonnet 4 (US)",
4393
+ "description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
4394
+ "family": "claude-sonnet",
4395
+ "attachment": true,
4396
+ "reasoning": true,
4397
+ "reasoning_options": [
4398
+ {
4399
+ "type": "toggle"
4400
+ },
4401
+ {
4402
+ "type": "budget_tokens",
4403
+ "min": 1024
4404
+ }
4405
+ ],
4406
+ "tool_call": true,
4407
+ "temperature": true,
4408
+ "knowledge": "2025-03-31",
4409
+ "release_date": "2025-05-22",
4410
+ "last_updated": "2025-05-22",
4411
+ "modalities": {
4412
+ "input": [
4413
+ "text",
4414
+ "image",
4415
+ "pdf"
4416
+ ],
4417
+ "output": [
4418
+ "text"
4419
+ ]
4420
+ },
4421
+ "open_weights": false,
4422
+ "limit": {
4423
+ "context": 200000,
4424
+ "output": 64000
4425
+ },
4426
+ "status": "deprecated",
4427
+ "cost": {
4428
+ "input": 3,
4429
+ "output": 15,
4430
+ "cache_read": 0.3,
4431
+ "cache_write": 3.75
4432
+ }
4433
+ },
3660
4434
  "moonshot.kimi-k2-thinking": {
3661
4435
  "id": "moonshot.kimi-k2-thinking",
3662
4436
  "name": "Kimi K2 Thinking",
3663
- "description": "Kimi reasoning model for long-horizon research, planning, and tool use",
4437
+ "description": "Thinking Kimi model for slower research passes, planning, and hard technical questions",
3664
4438
  "family": "kimi-thinking",
3665
4439
  "attachment": false,
3666
4440
  "reasoning": true,
@@ -3669,7 +4443,8 @@
3669
4443
  "interleaved": true,
3670
4444
  "structured_output": true,
3671
4445
  "temperature": true,
3672
- "release_date": "2025-12-02",
4446
+ "knowledge": "2024-08",
4447
+ "release_date": "2025-11-06",
3673
4448
  "last_updated": "2025-12-02",
3674
4449
  "modalities": {
3675
4450
  "input": [
@@ -3697,6 +4472,9 @@
3697
4472
  "attachment": true,
3698
4473
  "reasoning": true,
3699
4474
  "reasoning_options": [
4475
+ {
4476
+ "type": "toggle"
4477
+ },
3700
4478
  {
3701
4479
  "type": "budget_tokens",
3702
4480
  "min": 1024
@@ -3733,12 +4511,12 @@
3733
4511
  "deepseek.r1-v1:0": {
3734
4512
  "id": "deepseek.r1-v1:0",
3735
4513
  "name": "DeepSeek-R1",
3736
- "description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
4514
+ "description": "Classic open reasoning model for transparent math, coding, and deliberate problem solving",
3737
4515
  "family": "deepseek-thinking",
3738
4516
  "attachment": false,
3739
4517
  "reasoning": true,
3740
4518
  "reasoning_options": [],
3741
- "tool_call": true,
4519
+ "tool_call": false,
3742
4520
  "temperature": true,
3743
4521
  "knowledge": "2024-07",
3744
4522
  "release_date": "2025-01-20",
@@ -3751,7 +4529,7 @@
3751
4529
  "text"
3752
4530
  ]
3753
4531
  },
3754
- "open_weights": false,
4532
+ "open_weights": true,
3755
4533
  "limit": {
3756
4534
  "context": 128000,
3757
4535
  "output": 32768
@@ -3764,40 +4542,86 @@
3764
4542
  "mistral.magistral-small-2509": {
3765
4543
  "id": "mistral.magistral-small-2509",
3766
4544
  "name": "Magistral Small 1.2",
3767
- "description": "Mistral reasoning model for transparent analysis, math, and complex decisions",
4545
+ "description": "Open multimodal reasoning model for transparent analysis of text and images",
3768
4546
  "family": "magistral",
3769
- "attachment": false,
4547
+ "attachment": true,
3770
4548
  "reasoning": true,
3771
4549
  "reasoning_options": [],
3772
4550
  "tool_call": true,
3773
4551
  "structured_output": true,
3774
4552
  "temperature": true,
3775
- "release_date": "2025-12-02",
3776
- "last_updated": "2025-12-02",
4553
+ "release_date": "2025-09-18",
4554
+ "last_updated": "2025-09-18",
4555
+ "modalities": {
4556
+ "input": [
4557
+ "text",
4558
+ "image"
4559
+ ],
4560
+ "output": [
4561
+ "text"
4562
+ ]
4563
+ },
4564
+ "open_weights": true,
4565
+ "limit": {
4566
+ "context": 128000,
4567
+ "output": 40000
4568
+ },
4569
+ "cost": {
4570
+ "input": 0.5,
4571
+ "output": 1.5
4572
+ }
4573
+ },
4574
+ "us.anthropic.claude-fable-5": {
4575
+ "id": "us.anthropic.claude-fable-5",
4576
+ "name": "Claude Fable 5 (US)",
4577
+ "description": "Claude model for creative writing, analysis, and controlled agent workflows",
4578
+ "family": "claude-fable",
4579
+ "attachment": true,
4580
+ "reasoning": true,
4581
+ "reasoning_options": [
4582
+ {
4583
+ "type": "effort",
4584
+ "values": [
4585
+ "low",
4586
+ "medium",
4587
+ "high",
4588
+ "xhigh",
4589
+ "max"
4590
+ ]
4591
+ }
4592
+ ],
4593
+ "tool_call": true,
4594
+ "temperature": false,
4595
+ "knowledge": "2026-01-31",
4596
+ "release_date": "2026-06-09",
4597
+ "last_updated": "2026-06-09",
3777
4598
  "modalities": {
3778
4599
  "input": [
3779
4600
  "text",
3780
- "image"
4601
+ "image",
4602
+ "pdf"
3781
4603
  ],
3782
4604
  "output": [
3783
4605
  "text"
3784
4606
  ]
3785
4607
  },
3786
- "open_weights": true,
4608
+ "open_weights": false,
3787
4609
  "limit": {
3788
- "context": 128000,
3789
- "output": 40000
4610
+ "context": 1000000,
4611
+ "output": 128000
3790
4612
  },
3791
4613
  "cost": {
3792
- "input": 0.5,
3793
- "output": 1.5
4614
+ "input": 11,
4615
+ "output": 55,
4616
+ "cache_read": 1.1,
4617
+ "cache_write": 13.75
3794
4618
  }
3795
4619
  },
3796
- "us.anthropic.claude-fable-5": {
3797
- "id": "us.anthropic.claude-fable-5",
3798
- "name": "Claude Fable 5 (US)",
3799
- "description": "Claude model for creative writing, analysis, and controlled agent workflows",
3800
- "family": "claude-fable",
4620
+ "anthropic.claude-opus-5-5": {
4621
+ "id": "anthropic.claude-opus-5-5",
4622
+ "name": "Claude Opus 5.5",
4623
+ "description": "Claude model for long-running agentic coding and knowledge work",
4624
+ "family": "claude-opus",
3801
4625
  "attachment": true,
3802
4626
  "reasoning": true,
3803
4627
  "reasoning_options": [
@@ -3814,9 +4638,9 @@
3814
4638
  ],
3815
4639
  "tool_call": true,
3816
4640
  "temperature": false,
3817
- "knowledge": "2026-01-31",
3818
- "release_date": "2026-06-09",
3819
- "last_updated": "2026-06-09",
4641
+ "knowledge": "2026-06",
4642
+ "release_date": "2026-09-22",
4643
+ "last_updated": "2026-09-22",
3820
4644
  "modalities": {
3821
4645
  "input": [
3822
4646
  "text",
@@ -3833,10 +4657,10 @@
3833
4657
  "output": 128000
3834
4658
  },
3835
4659
  "cost": {
3836
- "input": 10,
3837
- "output": 50,
3838
- "cache_read": 1,
3839
- "cache_write": 12.5
4660
+ "input": 4,
4661
+ "output": 20,
4662
+ "cache_read": 0.2,
4663
+ "cache_write": 5
3840
4664
  }
3841
4665
  },
3842
4666
  "eu.anthropic.claude-fable-5": {
@@ -3913,8 +4737,7 @@
3913
4737
  "modalities": {
3914
4738
  "input": [
3915
4739
  "text",
3916
- "image",
3917
- "pdf"
4740
+ "image"
3918
4741
  ],
3919
4742
  "output": [
3920
4743
  "text"
@@ -3951,6 +4774,72 @@
3951
4774
  }
3952
4775
  }
3953
4776
  },
4777
+ "global.openai.gpt-6-luna": {
4778
+ "id": "global.openai.gpt-6-luna",
4779
+ "name": "GPT-6 Luna (Global)",
4780
+ "description": "OpenAI's most efficient model for focused, high-volume tasks",
4781
+ "family": "gpt-luna",
4782
+ "attachment": true,
4783
+ "reasoning": true,
4784
+ "reasoning_options": [
4785
+ {
4786
+ "type": "effort",
4787
+ "values": [
4788
+ "none",
4789
+ "low",
4790
+ "medium",
4791
+ "high",
4792
+ "xhigh",
4793
+ "max"
4794
+ ]
4795
+ }
4796
+ ],
4797
+ "tool_call": true,
4798
+ "structured_output": true,
4799
+ "temperature": false,
4800
+ "knowledge": "2026-05-18",
4801
+ "release_date": "2026-09-22",
4802
+ "last_updated": "2026-09-22",
4803
+ "modalities": {
4804
+ "input": [
4805
+ "text",
4806
+ "image"
4807
+ ],
4808
+ "output": [
4809
+ "text"
4810
+ ]
4811
+ },
4812
+ "open_weights": false,
4813
+ "limit": {
4814
+ "context": 1050000,
4815
+ "input": 922000,
4816
+ "output": 128000
4817
+ },
4818
+ "cost": {
4819
+ "input": 0.1,
4820
+ "output": 0.5,
4821
+ "cache_read": 0.01,
4822
+ "cache_write": 0.125,
4823
+ "tiers": [
4824
+ {
4825
+ "input": 0.2,
4826
+ "output": 0.75,
4827
+ "cache_read": 0.02,
4828
+ "cache_write": 0.25,
4829
+ "tier": {
4830
+ "type": "context",
4831
+ "size": 272000
4832
+ }
4833
+ }
4834
+ ],
4835
+ "context_over_200k": {
4836
+ "input": 0.2,
4837
+ "output": 0.75,
4838
+ "cache_read": 0.02,
4839
+ "cache_write": 0.25
4840
+ }
4841
+ }
4842
+ },
3954
4843
  "us.anthropic.claude-fable-5-1": {
3955
4844
  "id": "us.anthropic.claude-fable-5-1",
3956
4845
  "name": "Claude Fable 5.1 (US)",
@@ -4000,7 +4889,7 @@
4000
4889
  "meta.llama4-scout-17b-instruct-v1:0": {
4001
4890
  "id": "meta.llama4-scout-17b-instruct-v1:0",
4002
4891
  "name": "Llama 4 Scout 17B Instruct",
4003
- "description": "Open multimodal Llama model for long-context analysis and efficient agents",
4892
+ "description": "Open Llama with long-context vision for efficient multimodal agents",
4004
4893
  "family": "llama",
4005
4894
  "attachment": true,
4006
4895
  "reasoning": false,
@@ -4020,8 +4909,8 @@
4020
4909
  },
4021
4910
  "open_weights": true,
4022
4911
  "limit": {
4023
- "context": 3500000,
4024
- "output": 16384
4912
+ "context": 10000000,
4913
+ "output": 8192
4025
4914
  },
4026
4915
  "cost": {
4027
4916
  "input": 0.17,
@@ -4050,7 +4939,8 @@
4050
4939
  ],
4051
4940
  "tool_call": true,
4052
4941
  "temperature": true,
4053
- "release_date": "2025-12-01",
4942
+ "knowledge": "2025-10",
4943
+ "release_date": "2025-12-02",
4054
4944
  "last_updated": "2025-12-01",
4055
4945
  "modalities": {
4056
4946
  "input": [
@@ -4066,27 +4956,28 @@
4066
4956
  "open_weights": false,
4067
4957
  "limit": {
4068
4958
  "context": 1000000,
4069
- "output": 64000
4959
+ "output": 65535
4070
4960
  },
4071
4961
  "cost": {
4072
4962
  "input": 0.396,
4073
4963
  "output": 3.311,
4074
- "cache_read": 0.099
4964
+ "cache_read": 0.099,
4965
+ "cache_write": 0.396
4075
4966
  }
4076
4967
  },
4077
4968
  "google.gemma-3-27b-it": {
4078
4969
  "id": "google.gemma-3-27b-it",
4079
- "name": "Google Gemma 3 27B Instruct",
4080
- "description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
4970
+ "name": "Gemma 3 27B IT",
4971
+ "description": "Largest open Gemma 3 instruction model for multilingual text generation and visual understanding",
4081
4972
  "family": "gemma",
4082
4973
  "attachment": true,
4083
4974
  "reasoning": false,
4084
- "tool_call": true,
4975
+ "tool_call": false,
4085
4976
  "structured_output": true,
4086
4977
  "temperature": true,
4087
- "knowledge": "2025-07",
4088
- "release_date": "2025-07-27",
4089
- "last_updated": "2025-07-27",
4978
+ "knowledge": "2024-08",
4979
+ "release_date": "2025-03-12",
4980
+ "last_updated": "2025-03-12",
4090
4981
  "modalities": {
4091
4982
  "input": [
4092
4983
  "text",
@@ -4102,8 +4993,8 @@
4102
4993
  "output": 8192
4103
4994
  },
4104
4995
  "cost": {
4105
- "input": 0.12,
4106
- "output": 0.2
4996
+ "input": 0.23,
4997
+ "output": 0.38
4107
4998
  }
4108
4999
  },
4109
5000
  "amazon.nova-micro-v1:0": {
@@ -4129,12 +5020,13 @@
4129
5020
  "open_weights": false,
4130
5021
  "limit": {
4131
5022
  "context": 128000,
4132
- "output": 8192
5023
+ "output": 10000
4133
5024
  },
4134
5025
  "cost": {
4135
5026
  "input": 0.035,
4136
5027
  "output": 0.14,
4137
- "cache_read": 0.00875
5028
+ "cache_read": 0.00875,
5029
+ "cache_write": 0.035
4138
5030
  }
4139
5031
  },
4140
5032
  "us.mistral.pixtral-large-2502-v1:0": {
@@ -4316,6 +5208,9 @@
4316
5208
  "attachment": true,
4317
5209
  "reasoning": true,
4318
5210
  "reasoning_options": [
5211
+ {
5212
+ "type": "toggle"
5213
+ },
4319
5214
  {
4320
5215
  "type": "budget_tokens",
4321
5216
  "min": 1024
@@ -4343,10 +5238,10 @@
4343
5238
  "output": 64000
4344
5239
  },
4345
5240
  "cost": {
4346
- "input": 1,
4347
- "output": 5,
4348
- "cache_read": 0.1,
4349
- "cache_write": 1.25
5241
+ "input": 1.1,
5242
+ "output": 5.5,
5243
+ "cache_read": 0.11,
5244
+ "cache_write": 1.375
4350
5245
  }
4351
5246
  },
4352
5247
  "eu.anthropic.claude-sonnet-4-6": {
@@ -4357,6 +5252,9 @@
4357
5252
  "attachment": true,
4358
5253
  "reasoning": true,
4359
5254
  "reasoning_options": [
5255
+ {
5256
+ "type": "toggle"
5257
+ },
4360
5258
  {
4361
5259
  "type": "effort",
4362
5260
  "values": [
@@ -4390,7 +5288,7 @@
4390
5288
  "open_weights": false,
4391
5289
  "limit": {
4392
5290
  "context": 1000000,
4393
- "output": 64000
5291
+ "output": 128000
4394
5292
  },
4395
5293
  "cost": {
4396
5294
  "input": 3.3,
@@ -4399,6 +5297,73 @@
4399
5297
  "cache_write": 4.125
4400
5298
  }
4401
5299
  },
5300
+ "in.openai.gpt-5.6-terra": {
5301
+ "id": "in.openai.gpt-5.6-terra",
5302
+ "name": "GPT-5.6 Terra (India)",
5303
+ "description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
5304
+ "family": "gpt-terra",
5305
+ "attachment": true,
5306
+ "reasoning": true,
5307
+ "reasoning_options": [
5308
+ {
5309
+ "type": "effort",
5310
+ "values": [
5311
+ "none",
5312
+ "low",
5313
+ "medium",
5314
+ "high",
5315
+ "xhigh",
5316
+ "max"
5317
+ ]
5318
+ }
5319
+ ],
5320
+ "tool_call": true,
5321
+ "structured_output": false,
5322
+ "temperature": false,
5323
+ "knowledge": "2026-02-16",
5324
+ "release_date": "2026-07-09",
5325
+ "last_updated": "2026-07-09",
5326
+ "modalities": {
5327
+ "input": [
5328
+ "text",
5329
+ "image",
5330
+ "pdf"
5331
+ ],
5332
+ "output": [
5333
+ "text"
5334
+ ]
5335
+ },
5336
+ "open_weights": false,
5337
+ "limit": {
5338
+ "context": 1050000,
5339
+ "input": 922000,
5340
+ "output": 128000
5341
+ },
5342
+ "cost": {
5343
+ "input": 2.2,
5344
+ "output": 13.2,
5345
+ "cache_read": 0.22,
5346
+ "cache_write": 2.75,
5347
+ "tiers": [
5348
+ {
5349
+ "input": 4.4,
5350
+ "output": 19.8,
5351
+ "cache_read": 0.44,
5352
+ "cache_write": 5.5,
5353
+ "tier": {
5354
+ "type": "context",
5355
+ "size": 272000
5356
+ }
5357
+ }
5358
+ ],
5359
+ "context_over_200k": {
5360
+ "input": 4.4,
5361
+ "output": 19.8,
5362
+ "cache_read": 0.44,
5363
+ "cache_write": 5.5
5364
+ }
5365
+ }
5366
+ },
4402
5367
  "jp.anthropic.claude-opus-4-8": {
4403
5368
  "id": "jp.anthropic.claude-opus-4-8",
4404
5369
  "name": "Claude Opus 4.8 (JP)",
@@ -4407,6 +5372,9 @@
4407
5372
  "attachment": true,
4408
5373
  "reasoning": true,
4409
5374
  "reasoning_options": [
5375
+ {
5376
+ "type": "toggle"
5377
+ },
4410
5378
  {
4411
5379
  "type": "effort",
4412
5380
  "values": [
@@ -4439,10 +5407,10 @@
4439
5407
  "output": 128000
4440
5408
  },
4441
5409
  "cost": {
4442
- "input": 5,
4443
- "output": 25,
4444
- "cache_read": 0.5,
4445
- "cache_write": 6.25
5410
+ "input": 5.5,
5411
+ "output": 27.5,
5412
+ "cache_read": 0.55,
5413
+ "cache_write": 6.875
4446
5414
  }
4447
5415
  },
4448
5416
  "eu.anthropic.claude-haiku-4-5-20251001-v1:0": {
@@ -4453,6 +5421,9 @@
4453
5421
  "attachment": true,
4454
5422
  "reasoning": true,
4455
5423
  "reasoning_options": [
5424
+ {
5425
+ "type": "toggle"
5426
+ },
4456
5427
  {
4457
5428
  "type": "budget_tokens",
4458
5429
  "min": 1024
@@ -4488,7 +5459,7 @@
4488
5459
  },
4489
5460
  "qwen.qwen3-next-80b-a3b": {
4490
5461
  "id": "qwen.qwen3-next-80b-a3b",
4491
- "name": "Qwen/Qwen3-Next-80B-A3B-Instruct",
5462
+ "name": "Qwen3-Next 80B-A3B Instruct",
4492
5463
  "description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
4493
5464
  "family": "qwen",
4494
5465
  "attachment": false,
@@ -4496,7 +5467,8 @@
4496
5467
  "tool_call": true,
4497
5468
  "structured_output": true,
4498
5469
  "temperature": true,
4499
- "release_date": "2025-09-18",
5470
+ "knowledge": "2025-04",
5471
+ "release_date": "2025-09-11",
4500
5472
  "last_updated": "2025-11-25",
4501
5473
  "modalities": {
4502
5474
  "input": [
@@ -4506,14 +5478,14 @@
4506
5478
  "text"
4507
5479
  ]
4508
5480
  },
4509
- "open_weights": false,
5481
+ "open_weights": true,
4510
5482
  "limit": {
4511
- "context": 262000,
5483
+ "context": 262144,
4512
5484
  "output": 262000
4513
5485
  },
4514
5486
  "cost": {
4515
- "input": 0.14,
4516
- "output": 1.4
5487
+ "input": 0.15,
5488
+ "output": 1.2
4517
5489
  }
4518
5490
  },
4519
5491
  "us.anthropic.claude-sonnet-5": {
@@ -4539,7 +5511,7 @@
4539
5511
  }
4540
5512
  ],
4541
5513
  "tool_call": true,
4542
- "structured_output": true,
5514
+ "structured_output": false,
4543
5515
  "temperature": false,
4544
5516
  "knowledge": "2026-01-31",
4545
5517
  "release_date": "2026-06-30",
@@ -4560,10 +5532,10 @@
4560
5532
  "output": 128000
4561
5533
  },
4562
5534
  "cost": {
4563
- "input": 2,
4564
- "output": 10,
4565
- "cache_read": 0.2,
4566
- "cache_write": 2.5
5535
+ "input": 2.2,
5536
+ "output": 11,
5537
+ "cache_read": 0.22,
5538
+ "cache_write": 2.75
4567
5539
  }
4568
5540
  },
4569
5541
  "anthropic.claude-sonnet-4-5-20250929-v1:0": {
@@ -4574,6 +5546,9 @@
4574
5546
  "attachment": true,
4575
5547
  "reasoning": true,
4576
5548
  "reasoning_options": [
5549
+ {
5550
+ "type": "toggle"
5551
+ },
4577
5552
  {
4578
5553
  "type": "budget_tokens",
4579
5554
  "min": 1024
@@ -4623,7 +5598,8 @@
4623
5598
  "input": [
4624
5599
  "text",
4625
5600
  "image",
4626
- "video"
5601
+ "video",
5602
+ "pdf"
4627
5603
  ],
4628
5604
  "output": [
4629
5605
  "text"
@@ -4632,22 +5608,92 @@
4632
5608
  "open_weights": false,
4633
5609
  "limit": {
4634
5610
  "context": 300000,
4635
- "output": 8192
5611
+ "output": 10000
5612
+ },
5613
+ "cost": {
5614
+ "input": 0.06,
5615
+ "output": 0.24,
5616
+ "cache_read": 0.015,
5617
+ "cache_write": 0.06
5618
+ }
5619
+ },
5620
+ "us.openai.gpt-6-sol": {
5621
+ "id": "us.openai.gpt-6-sol",
5622
+ "name": "GPT-6 Sol (US)",
5623
+ "description": "OpenAI model for complex coding and agentic workflows",
5624
+ "family": "gpt-sol",
5625
+ "attachment": true,
5626
+ "reasoning": true,
5627
+ "reasoning_options": [
5628
+ {
5629
+ "type": "effort",
5630
+ "values": [
5631
+ "none",
5632
+ "low",
5633
+ "medium",
5634
+ "high",
5635
+ "xhigh",
5636
+ "max"
5637
+ ]
5638
+ }
5639
+ ],
5640
+ "tool_call": true,
5641
+ "structured_output": true,
5642
+ "temperature": false,
5643
+ "knowledge": "2026-04-20",
5644
+ "release_date": "2026-09-22",
5645
+ "last_updated": "2026-09-22",
5646
+ "modalities": {
5647
+ "input": [
5648
+ "text",
5649
+ "image"
5650
+ ],
5651
+ "output": [
5652
+ "text"
5653
+ ]
5654
+ },
5655
+ "open_weights": false,
5656
+ "limit": {
5657
+ "context": 1050000,
5658
+ "input": 922000,
5659
+ "output": 128000
4636
5660
  },
4637
5661
  "cost": {
4638
- "input": 0.06,
4639
- "output": 0.24,
4640
- "cache_read": 0.015
5662
+ "input": 2.2,
5663
+ "output": 11,
5664
+ "cache_read": 0.22,
5665
+ "cache_write": 2.75,
5666
+ "tiers": [
5667
+ {
5668
+ "input": 4.4,
5669
+ "output": 16.5,
5670
+ "cache_read": 0.44,
5671
+ "cache_write": 5.5,
5672
+ "tier": {
5673
+ "type": "context",
5674
+ "size": 272000
5675
+ }
5676
+ }
5677
+ ],
5678
+ "context_over_200k": {
5679
+ "input": 4.4,
5680
+ "output": 16.5,
5681
+ "cache_read": 0.44,
5682
+ "cache_write": 5.5
5683
+ }
4641
5684
  }
4642
5685
  },
4643
5686
  "global.anthropic.claude-opus-4-7": {
4644
5687
  "id": "global.anthropic.claude-opus-4-7",
4645
5688
  "name": "Claude Opus 4.7 (Global)",
4646
- "description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
5689
+ "description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
4647
5690
  "family": "claude-opus",
4648
5691
  "attachment": true,
4649
5692
  "reasoning": true,
4650
5693
  "reasoning_options": [
5694
+ {
5695
+ "type": "toggle"
5696
+ },
4651
5697
  {
4652
5698
  "type": "effort",
4653
5699
  "values": [
@@ -4688,16 +5734,16 @@
4688
5734
  },
4689
5735
  "qwen.qwen3-coder-480b-a35b-v1:0": {
4690
5736
  "id": "qwen.qwen3-coder-480b-a35b-v1:0",
4691
- "name": "Qwen3 Coder 480B A35B Instruct",
4692
- "description": "Qwen coding model for software agents, repository edits, and code reasoning",
5737
+ "name": "Qwen3-Coder 480B-A35B Instruct",
5738
+ "description": "Open Qwen coding heavyweight for repository reasoning and agentic engineering",
4693
5739
  "family": "qwen",
4694
5740
  "attachment": false,
4695
5741
  "reasoning": false,
4696
5742
  "tool_call": true,
4697
5743
  "structured_output": true,
4698
5744
  "temperature": true,
4699
- "knowledge": "2024-04",
4700
- "release_date": "2025-09-18",
5745
+ "knowledge": "2025-04",
5746
+ "release_date": "2025-07-23",
4701
5747
  "last_updated": "2025-09-18",
4702
5748
  "modalities": {
4703
5749
  "input": [
@@ -4713,7 +5759,7 @@
4713
5759
  "output": 65536
4714
5760
  },
4715
5761
  "cost": {
4716
- "input": 0.22,
5762
+ "input": 0.45,
4717
5763
  "output": 1.8
4718
5764
  }
4719
5765
  },
@@ -4823,12 +5869,17 @@
4823
5869
  "zai.glm-4.7-flash": {
4824
5870
  "id": "zai.glm-4.7-flash",
4825
5871
  "name": "GLM-4.7-Flash",
4826
- "description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
5872
+ "description": "Budget GLM lane for fast coding help, routing, and everyday automation",
4827
5873
  "family": "glm-flash",
4828
5874
  "attachment": false,
4829
5875
  "reasoning": true,
4830
- "reasoning_options": [],
5876
+ "reasoning_options": [
5877
+ {
5878
+ "type": "toggle"
5879
+ }
5880
+ ],
4831
5881
  "tool_call": true,
5882
+ "interleaved": true,
4832
5883
  "structured_output": true,
4833
5884
  "temperature": true,
4834
5885
  "knowledge": "2025-04",
@@ -4855,14 +5906,15 @@
4855
5906
  "google.gemma-3-4b-it": {
4856
5907
  "id": "google.gemma-3-4b-it",
4857
5908
  "name": "Gemma 3 4B IT",
4858
- "description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
5909
+ "description": "Open multimodal Gemma instruction model for efficient text generation and image understanding",
4859
5910
  "family": "gemma",
4860
- "attachment": false,
5911
+ "attachment": true,
4861
5912
  "reasoning": false,
4862
- "tool_call": true,
5913
+ "tool_call": false,
4863
5914
  "temperature": true,
4864
- "release_date": "2024-12-01",
4865
- "last_updated": "2024-12-01",
5915
+ "knowledge": "2024-08",
5916
+ "release_date": "2025-03-12",
5917
+ "last_updated": "2025-03-12",
4866
5918
  "modalities": {
4867
5919
  "input": [
4868
5920
  "text",
@@ -4872,9 +5924,9 @@
4872
5924
  "text"
4873
5925
  ]
4874
5926
  },
4875
- "open_weights": false,
5927
+ "open_weights": true,
4876
5928
  "limit": {
4877
- "context": 128000,
5929
+ "context": 131072,
4878
5930
  "output": 4096
4879
5931
  },
4880
5932
  "cost": {
@@ -4911,8 +5963,7 @@
4911
5963
  "modalities": {
4912
5964
  "input": [
4913
5965
  "text",
4914
- "image",
4915
- "pdf"
5966
+ "image"
4916
5967
  ],
4917
5968
  "output": [
4918
5969
  "text"
@@ -4952,18 +6003,20 @@
4952
6003
  "zai.glm-5": {
4953
6004
  "id": "zai.glm-5",
4954
6005
  "name": "GLM-5",
4955
- "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
6006
+ "description": "General GLM flagship for coding, analysis, and tool-heavy engineering workflows",
4956
6007
  "family": "glm",
4957
6008
  "attachment": false,
4958
6009
  "reasoning": true,
4959
- "reasoning_options": [],
6010
+ "reasoning_options": [
6011
+ {
6012
+ "type": "toggle"
6013
+ }
6014
+ ],
4960
6015
  "tool_call": true,
4961
- "interleaved": {
4962
- "field": "reasoning_content"
4963
- },
6016
+ "interleaved": true,
4964
6017
  "structured_output": true,
4965
6018
  "temperature": true,
4966
- "release_date": "2026-03-18",
6019
+ "release_date": "2026-02-12",
4967
6020
  "last_updated": "2026-03-18",
4968
6021
  "modalities": {
4969
6022
  "input": [
@@ -4976,7 +6029,7 @@
4976
6029
  "open_weights": true,
4977
6030
  "limit": {
4978
6031
  "context": 202752,
4979
- "output": 101376
6032
+ "output": 131072
4980
6033
  },
4981
6034
  "cost": {
4982
6035
  "input": 1,
@@ -4989,7 +6042,17 @@
4989
6042
  "description": "Safety model for policy screening, moderation, and risk-aware routing workflows",
4990
6043
  "family": "gpt-oss",
4991
6044
  "attachment": false,
4992
- "reasoning": false,
6045
+ "reasoning": true,
6046
+ "reasoning_options": [
6047
+ {
6048
+ "type": "effort",
6049
+ "values": [
6050
+ "low",
6051
+ "medium",
6052
+ "high"
6053
+ ]
6054
+ }
6055
+ ],
4993
6056
  "tool_call": true,
4994
6057
  "structured_output": true,
4995
6058
  "temperature": true,
@@ -5016,15 +6079,16 @@
5016
6079
  "mistral.devstral-2-123b": {
5017
6080
  "id": "mistral.devstral-2-123b",
5018
6081
  "name": "Devstral 2 123B",
5019
- "description": "Mistral coding agent model for repository tasks and software engineering workflows",
6082
+ "description": "Mistral's coding-agent model for repository work, terminal tasks, and software fixes",
5020
6083
  "family": "devstral",
5021
6084
  "attachment": false,
5022
6085
  "reasoning": false,
5023
6086
  "tool_call": true,
5024
6087
  "structured_output": true,
5025
6088
  "temperature": true,
5026
- "release_date": "2026-02-17",
5027
- "last_updated": "2026-02-17",
6089
+ "knowledge": "2025-12",
6090
+ "release_date": "2025-12-09",
6091
+ "last_updated": "2025-12-09",
5028
6092
  "modalities": {
5029
6093
  "input": [
5030
6094
  "text"
@@ -5122,6 +6186,9 @@
5122
6186
  "attachment": true,
5123
6187
  "reasoning": true,
5124
6188
  "reasoning_options": [
6189
+ {
6190
+ "type": "toggle"
6191
+ },
5125
6192
  {
5126
6193
  "type": "budget_tokens",
5127
6194
  "min": 1024
@@ -5163,6 +6230,9 @@
5163
6230
  "attachment": true,
5164
6231
  "reasoning": true,
5165
6232
  "reasoning_options": [
6233
+ {
6234
+ "type": "toggle"
6235
+ },
5166
6236
  {
5167
6237
  "type": "effort",
5168
6238
  "values": [
@@ -5196,27 +6266,27 @@
5196
6266
  "open_weights": false,
5197
6267
  "limit": {
5198
6268
  "context": 1000000,
5199
- "output": 64000
6269
+ "output": 128000
5200
6270
  },
5201
6271
  "cost": {
5202
- "input": 3,
5203
- "output": 15,
5204
- "cache_read": 0.3,
5205
- "cache_write": 3.75
6272
+ "input": 3.3,
6273
+ "output": 16.5,
6274
+ "cache_read": 0.33,
6275
+ "cache_write": 4.125
5206
6276
  }
5207
6277
  },
5208
6278
  "mistral.voxtral-small-24b-2507": {
5209
6279
  "id": "mistral.voxtral-small-24b-2507",
5210
6280
  "name": "Voxtral Small 24B 2507",
5211
- "description": "Efficient Mistral model for fast chat, extraction, and production assistants",
5212
- "family": "mistral",
6281
+ "description": "Open audio-language model for speech transcription, audio understanding, and voice-driven tool use",
6282
+ "family": "voxtral",
5213
6283
  "attachment": true,
5214
6284
  "reasoning": false,
5215
6285
  "tool_call": true,
5216
6286
  "structured_output": true,
5217
6287
  "temperature": true,
5218
- "release_date": "2025-07-01",
5219
- "last_updated": "2025-07-01",
6288
+ "release_date": "2025-07-15",
6289
+ "last_updated": "2025-07-15",
5220
6290
  "modalities": {
5221
6291
  "input": [
5222
6292
  "text",
@@ -5228,18 +6298,18 @@
5228
6298
  },
5229
6299
  "open_weights": true,
5230
6300
  "limit": {
5231
- "context": 32000,
6301
+ "context": 32768,
5232
6302
  "output": 8192
5233
6303
  },
5234
6304
  "cost": {
5235
- "input": 0.15,
5236
- "output": 0.35
6305
+ "input": 0.1,
6306
+ "output": 0.3
5237
6307
  }
5238
6308
  },
5239
6309
  "openai.gpt-oss-20b": {
5240
6310
  "id": "openai.gpt-oss-20b",
5241
6311
  "name": "gpt-oss-20b",
5242
- "description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
6312
+ "description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
5243
6313
  "family": "gpt-oss",
5244
6314
  "attachment": false,
5245
6315
  "reasoning": true,
@@ -5284,7 +6354,7 @@
5284
6354
  "meta.llama4-maverick-17b-instruct-v1:0": {
5285
6355
  "id": "meta.llama4-maverick-17b-instruct-v1:0",
5286
6356
  "name": "Llama 4 Maverick 17B Instruct",
5287
- "description": "Open multimodal Llama model for strong reasoning and fast responses",
6357
+ "description": "Open multimodal Llama for strong reasoning with efficient everyday serving",
5288
6358
  "family": "llama",
5289
6359
  "attachment": true,
5290
6360
  "reasoning": false,
@@ -5305,25 +6375,73 @@
5305
6375
  "open_weights": true,
5306
6376
  "limit": {
5307
6377
  "context": 1000000,
5308
- "output": 16384
6378
+ "output": 8192
5309
6379
  },
5310
6380
  "cost": {
5311
6381
  "input": 0.24,
5312
6382
  "output": 0.97
5313
6383
  }
5314
6384
  },
6385
+ "eu.anthropic.claude-opus-5-5": {
6386
+ "id": "eu.anthropic.claude-opus-5-5",
6387
+ "name": "Claude Opus 5.5 (EU)",
6388
+ "description": "Claude model for long-running agentic coding and knowledge work",
6389
+ "family": "claude-opus",
6390
+ "attachment": true,
6391
+ "reasoning": true,
6392
+ "reasoning_options": [
6393
+ {
6394
+ "type": "effort",
6395
+ "values": [
6396
+ "low",
6397
+ "medium",
6398
+ "high",
6399
+ "xhigh",
6400
+ "max"
6401
+ ]
6402
+ }
6403
+ ],
6404
+ "tool_call": true,
6405
+ "temperature": false,
6406
+ "knowledge": "2026-06",
6407
+ "release_date": "2026-09-22",
6408
+ "last_updated": "2026-09-22",
6409
+ "modalities": {
6410
+ "input": [
6411
+ "text",
6412
+ "image",
6413
+ "pdf"
6414
+ ],
6415
+ "output": [
6416
+ "text"
6417
+ ]
6418
+ },
6419
+ "open_weights": false,
6420
+ "limit": {
6421
+ "context": 1000000,
6422
+ "output": 128000
6423
+ },
6424
+ "cost": {
6425
+ "input": 4.4,
6426
+ "output": 22,
6427
+ "cache_read": 0.22,
6428
+ "cache_write": 5.5
6429
+ }
6430
+ },
5315
6431
  "zai.glm-4.7": {
5316
6432
  "id": "zai.glm-4.7",
5317
6433
  "name": "GLM-4.7",
5318
- "description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
6434
+ "description": "Mature GLM model for dependable coding, reasoning, and structured agent tasks",
5319
6435
  "family": "glm",
5320
6436
  "attachment": false,
5321
6437
  "reasoning": true,
5322
- "reasoning_options": [],
6438
+ "reasoning_options": [
6439
+ {
6440
+ "type": "toggle"
6441
+ }
6442
+ ],
5323
6443
  "tool_call": true,
5324
- "interleaved": {
5325
- "field": "reasoning_content"
5326
- },
6444
+ "interleaved": true,
5327
6445
  "structured_output": true,
5328
6446
  "temperature": true,
5329
6447
  "knowledge": "2025-04",
@@ -5363,7 +6481,8 @@
5363
6481
  "input": [
5364
6482
  "text",
5365
6483
  "image",
5366
- "video"
6484
+ "video",
6485
+ "pdf"
5367
6486
  ],
5368
6487
  "output": [
5369
6488
  "text"
@@ -5372,12 +6491,13 @@
5372
6491
  "open_weights": false,
5373
6492
  "limit": {
5374
6493
  "context": 300000,
5375
- "output": 8192
6494
+ "output": 10000
5376
6495
  },
5377
6496
  "cost": {
5378
6497
  "input": 0.064,
5379
6498
  "output": 0.256,
5380
- "cache_read": 0.016
6499
+ "cache_read": 0.016,
6500
+ "cache_write": 0.064
5381
6501
  }
5382
6502
  },
5383
6503
  "us.anthropic.claude-opus-4-5-20251101-v1:0": {
@@ -5388,6 +6508,9 @@
5388
6508
  "attachment": true,
5389
6509
  "reasoning": true,
5390
6510
  "reasoning_options": [
6511
+ {
6512
+ "type": "toggle"
6513
+ },
5391
6514
  {
5392
6515
  "type": "effort",
5393
6516
  "values": [
@@ -5404,9 +6527,9 @@
5404
6527
  "tool_call": true,
5405
6528
  "structured_output": true,
5406
6529
  "temperature": true,
5407
- "knowledge": "2025-03-31",
5408
- "release_date": "2025-11-24",
5409
- "last_updated": "2025-08-01",
6530
+ "knowledge": "2025-05",
6531
+ "release_date": "2025-11-01",
6532
+ "last_updated": "2025-11-01",
5410
6533
  "modalities": {
5411
6534
  "input": [
5412
6535
  "text",
@@ -5423,10 +6546,10 @@
5423
6546
  "output": 64000
5424
6547
  },
5425
6548
  "cost": {
5426
- "input": 5,
5427
- "output": 25,
5428
- "cache_read": 0.5,
5429
- "cache_write": 6.25
6549
+ "input": 5.5,
6550
+ "output": 27.5,
6551
+ "cache_read": 0.55,
6552
+ "cache_write": 6.875
5430
6553
  }
5431
6554
  },
5432
6555
  "au.anthropic.claude-opus-4-7": {
@@ -5437,6 +6560,9 @@
5437
6560
  "attachment": true,
5438
6561
  "reasoning": true,
5439
6562
  "reasoning_options": [
6563
+ {
6564
+ "type": "toggle"
6565
+ },
5440
6566
  {
5441
6567
  "type": "effort",
5442
6568
  "values": [
@@ -5483,6 +6609,9 @@
5483
6609
  "attachment": true,
5484
6610
  "reasoning": true,
5485
6611
  "reasoning_options": [
6612
+ {
6613
+ "type": "toggle"
6614
+ },
5486
6615
  {
5487
6616
  "type": "effort",
5488
6617
  "values": [
@@ -5516,13 +6645,13 @@
5516
6645
  "open_weights": false,
5517
6646
  "limit": {
5518
6647
  "context": 1000000,
5519
- "output": 64000
6648
+ "output": 128000
5520
6649
  },
5521
6650
  "cost": {
5522
- "input": 3,
5523
- "output": 15,
5524
- "cache_read": 0.3,
5525
- "cache_write": 3.75
6651
+ "input": 3.3,
6652
+ "output": 16.5,
6653
+ "cache_read": 0.33,
6654
+ "cache_write": 4.125
5526
6655
  }
5527
6656
  },
5528
6657
  "us.anthropic.claude-sonnet-4-5-20250929-v1:0": {
@@ -5534,16 +6663,111 @@
5534
6663
  "reasoning": true,
5535
6664
  "reasoning_options": [
5536
6665
  {
5537
- "type": "budget_tokens",
5538
- "min": 1024
6666
+ "type": "toggle"
6667
+ },
6668
+ {
6669
+ "type": "budget_tokens",
6670
+ "min": 1024
6671
+ }
6672
+ ],
6673
+ "tool_call": true,
6674
+ "structured_output": true,
6675
+ "temperature": true,
6676
+ "knowledge": "2025-07-31",
6677
+ "release_date": "2025-09-29",
6678
+ "last_updated": "2025-09-29",
6679
+ "modalities": {
6680
+ "input": [
6681
+ "text",
6682
+ "image",
6683
+ "pdf"
6684
+ ],
6685
+ "output": [
6686
+ "text"
6687
+ ]
6688
+ },
6689
+ "open_weights": false,
6690
+ "limit": {
6691
+ "context": 200000,
6692
+ "output": 64000
6693
+ },
6694
+ "cost": {
6695
+ "input": 3.3,
6696
+ "output": 16.5,
6697
+ "cache_read": 0.33,
6698
+ "cache_write": 4.125
6699
+ }
6700
+ },
6701
+ "global.anthropic.claude-opus-5-5": {
6702
+ "id": "global.anthropic.claude-opus-5-5",
6703
+ "name": "Claude Opus 5.5 (Global)",
6704
+ "description": "Claude model for long-running agentic coding and knowledge work",
6705
+ "family": "claude-opus",
6706
+ "attachment": true,
6707
+ "reasoning": true,
6708
+ "reasoning_options": [
6709
+ {
6710
+ "type": "effort",
6711
+ "values": [
6712
+ "low",
6713
+ "medium",
6714
+ "high",
6715
+ "xhigh",
6716
+ "max"
6717
+ ]
6718
+ }
6719
+ ],
6720
+ "tool_call": true,
6721
+ "temperature": false,
6722
+ "knowledge": "2026-06",
6723
+ "release_date": "2026-09-22",
6724
+ "last_updated": "2026-09-22",
6725
+ "modalities": {
6726
+ "input": [
6727
+ "text",
6728
+ "image",
6729
+ "pdf"
6730
+ ],
6731
+ "output": [
6732
+ "text"
6733
+ ]
6734
+ },
6735
+ "open_weights": false,
6736
+ "limit": {
6737
+ "context": 1000000,
6738
+ "output": 128000
6739
+ },
6740
+ "cost": {
6741
+ "input": 4,
6742
+ "output": 20,
6743
+ "cache_read": 0.2,
6744
+ "cache_write": 5
6745
+ }
6746
+ },
6747
+ "jp.anthropic.claude-opus-5-5": {
6748
+ "id": "jp.anthropic.claude-opus-5-5",
6749
+ "name": "Claude Opus 5.5 (JP)",
6750
+ "description": "Claude model for long-running agentic coding and knowledge work",
6751
+ "family": "claude-opus",
6752
+ "attachment": true,
6753
+ "reasoning": true,
6754
+ "reasoning_options": [
6755
+ {
6756
+ "type": "effort",
6757
+ "values": [
6758
+ "low",
6759
+ "medium",
6760
+ "high",
6761
+ "xhigh",
6762
+ "max"
6763
+ ]
5539
6764
  }
5540
6765
  ],
5541
6766
  "tool_call": true,
5542
- "structured_output": true,
5543
- "temperature": true,
5544
- "knowledge": "2025-07-31",
5545
- "release_date": "2025-09-29",
5546
- "last_updated": "2025-09-29",
6767
+ "temperature": false,
6768
+ "knowledge": "2026-06",
6769
+ "release_date": "2026-09-22",
6770
+ "last_updated": "2026-09-22",
5547
6771
  "modalities": {
5548
6772
  "input": [
5549
6773
  "text",
@@ -5556,14 +6780,14 @@
5556
6780
  },
5557
6781
  "open_weights": false,
5558
6782
  "limit": {
5559
- "context": 200000,
5560
- "output": 64000
6783
+ "context": 1000000,
6784
+ "output": 128000
5561
6785
  },
5562
6786
  "cost": {
5563
- "input": 3,
5564
- "output": 15,
5565
- "cache_read": 0.3,
5566
- "cache_write": 3.75
6787
+ "input": 4.4,
6788
+ "output": 22,
6789
+ "cache_read": 0.22,
6790
+ "cache_write": 5.5
5567
6791
  }
5568
6792
  },
5569
6793
  "us.deepseek.r1-v1:0": {
@@ -5574,7 +6798,7 @@
5574
6798
  "attachment": false,
5575
6799
  "reasoning": true,
5576
6800
  "reasoning_options": [],
5577
- "tool_call": true,
6801
+ "tool_call": false,
5578
6802
  "temperature": true,
5579
6803
  "knowledge": "2024-07",
5580
6804
  "release_date": "2025-01-20",
@@ -5597,6 +6821,52 @@
5597
6821
  "output": 5.4
5598
6822
  }
5599
6823
  },
6824
+ "us.anthropic.claude-opus-5-5": {
6825
+ "id": "us.anthropic.claude-opus-5-5",
6826
+ "name": "Claude Opus 5.5 (US)",
6827
+ "description": "Claude model for long-running agentic coding and knowledge work",
6828
+ "family": "claude-opus",
6829
+ "attachment": true,
6830
+ "reasoning": true,
6831
+ "reasoning_options": [
6832
+ {
6833
+ "type": "effort",
6834
+ "values": [
6835
+ "low",
6836
+ "medium",
6837
+ "high",
6838
+ "xhigh",
6839
+ "max"
6840
+ ]
6841
+ }
6842
+ ],
6843
+ "tool_call": true,
6844
+ "temperature": false,
6845
+ "knowledge": "2026-06",
6846
+ "release_date": "2026-09-22",
6847
+ "last_updated": "2026-09-22",
6848
+ "modalities": {
6849
+ "input": [
6850
+ "text",
6851
+ "image",
6852
+ "pdf"
6853
+ ],
6854
+ "output": [
6855
+ "text"
6856
+ ]
6857
+ },
6858
+ "open_weights": false,
6859
+ "limit": {
6860
+ "context": 1000000,
6861
+ "output": 128000
6862
+ },
6863
+ "cost": {
6864
+ "input": 4.4,
6865
+ "output": 22,
6866
+ "cache_read": 0.22,
6867
+ "cache_write": 5.5
6868
+ }
6869
+ },
5600
6870
  "us.anthropic.claude-opus-4-8": {
5601
6871
  "id": "us.anthropic.claude-opus-4-8",
5602
6872
  "name": "Claude Opus 4.8 (US)",
@@ -5605,6 +6875,9 @@
5605
6875
  "attachment": true,
5606
6876
  "reasoning": true,
5607
6877
  "reasoning_options": [
6878
+ {
6879
+ "type": "toggle"
6880
+ },
5608
6881
  {
5609
6882
  "type": "effort",
5610
6883
  "values": [
@@ -5637,10 +6910,10 @@
5637
6910
  "output": 128000
5638
6911
  },
5639
6912
  "cost": {
5640
- "input": 5,
5641
- "output": 25,
5642
- "cache_read": 0.5,
5643
- "cache_write": 6.25
6913
+ "input": 5.5,
6914
+ "output": 27.5,
6915
+ "cache_read": 0.55,
6916
+ "cache_write": 6.875
5644
6917
  }
5645
6918
  },
5646
6919
  "au.anthropic.claude-opus-5": {
@@ -5651,6 +6924,9 @@
5651
6924
  "attachment": true,
5652
6925
  "reasoning": true,
5653
6926
  "reasoning_options": [
6927
+ {
6928
+ "type": "toggle"
6929
+ },
5654
6930
  {
5655
6931
  "type": "effort",
5656
6932
  "values": [
@@ -5683,10 +6959,10 @@
5683
6959
  "output": 128000
5684
6960
  },
5685
6961
  "cost": {
5686
- "input": 5,
5687
- "output": 25,
5688
- "cache_read": 0.5,
5689
- "cache_write": 6.25
6962
+ "input": 5.5,
6963
+ "output": 27.5,
6964
+ "cache_read": 0.55,
6965
+ "cache_write": 6.875
5690
6966
  }
5691
6967
  },
5692
6968
  "anthropic.claude-opus-4-1-20250805-v1:0": {
@@ -5697,6 +6973,9 @@
5697
6973
  "attachment": true,
5698
6974
  "reasoning": true,
5699
6975
  "reasoning_options": [
6976
+ {
6977
+ "type": "toggle"
6978
+ },
5700
6979
  {
5701
6980
  "type": "budget_tokens",
5702
6981
  "min": 1024
@@ -5730,6 +7009,50 @@
5730
7009
  "cache_write": 18.75
5731
7010
  }
5732
7011
  },
7012
+ "apac.anthropic.claude-sonnet-4-20250514-v1:0": {
7013
+ "id": "apac.anthropic.claude-sonnet-4-20250514-v1:0",
7014
+ "name": "Claude Sonnet 4 (APAC)",
7015
+ "description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
7016
+ "family": "claude-sonnet",
7017
+ "attachment": true,
7018
+ "reasoning": true,
7019
+ "reasoning_options": [
7020
+ {
7021
+ "type": "toggle"
7022
+ },
7023
+ {
7024
+ "type": "budget_tokens",
7025
+ "min": 1024
7026
+ }
7027
+ ],
7028
+ "tool_call": true,
7029
+ "temperature": true,
7030
+ "knowledge": "2025-03-31",
7031
+ "release_date": "2025-05-22",
7032
+ "last_updated": "2025-05-22",
7033
+ "modalities": {
7034
+ "input": [
7035
+ "text",
7036
+ "image",
7037
+ "pdf"
7038
+ ],
7039
+ "output": [
7040
+ "text"
7041
+ ]
7042
+ },
7043
+ "open_weights": false,
7044
+ "limit": {
7045
+ "context": 200000,
7046
+ "output": 64000
7047
+ },
7048
+ "status": "deprecated",
7049
+ "cost": {
7050
+ "input": 3,
7051
+ "output": 15,
7052
+ "cache_read": 0.3,
7053
+ "cache_write": 3.75
7054
+ }
7055
+ },
5733
7056
  "jp.anthropic.claude-sonnet-5": {
5734
7057
  "id": "jp.anthropic.claude-sonnet-5",
5735
7058
  "name": "Claude Sonnet 5 (JP)",
@@ -5753,7 +7076,7 @@
5753
7076
  }
5754
7077
  ],
5755
7078
  "tool_call": true,
5756
- "structured_output": true,
7079
+ "structured_output": false,
5757
7080
  "temperature": false,
5758
7081
  "knowledge": "2026-01-31",
5759
7082
  "release_date": "2026-06-30",
@@ -5774,10 +7097,10 @@
5774
7097
  "output": 128000
5775
7098
  },
5776
7099
  "cost": {
5777
- "input": 2,
5778
- "output": 10,
5779
- "cache_read": 0.2,
5780
- "cache_write": 2.5
7100
+ "input": 2.2,
7101
+ "output": 11,
7102
+ "cache_read": 0.22,
7103
+ "cache_write": 2.75
5781
7104
  }
5782
7105
  },
5783
7106
  "au.anthropic.claude-sonnet-5": {
@@ -5803,7 +7126,7 @@
5803
7126
  }
5804
7127
  ],
5805
7128
  "tool_call": true,
5806
- "structured_output": true,
7129
+ "structured_output": false,
5807
7130
  "temperature": false,
5808
7131
  "knowledge": "2026-01-31",
5809
7132
  "release_date": "2026-06-30",
@@ -5824,10 +7147,10 @@
5824
7147
  "output": 128000
5825
7148
  },
5826
7149
  "cost": {
5827
- "input": 2,
5828
- "output": 10,
5829
- "cache_read": 0.2,
5830
- "cache_write": 2.5
7150
+ "input": 2.2,
7151
+ "output": 11,
7152
+ "cache_read": 0.22,
7153
+ "cache_write": 2.75
5831
7154
  }
5832
7155
  },
5833
7156
  "xai.grok-4.3": {
@@ -5881,7 +7204,7 @@
5881
7204
  "openai.gpt-oss-120b": {
5882
7205
  "id": "openai.gpt-oss-120b",
5883
7206
  "name": "gpt-oss-120b",
5884
- "description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
7207
+ "description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
5885
7208
  "family": "gpt-oss",
5886
7209
  "attachment": false,
5887
7210
  "reasoning": true,
@@ -6007,6 +7330,9 @@
6007
7330
  "attachment": true,
6008
7331
  "reasoning": true,
6009
7332
  "reasoning_options": [
7333
+ {
7334
+ "type": "toggle"
7335
+ },
6010
7336
  {
6011
7337
  "type": "budget_tokens",
6012
7338
  "min": 1024
@@ -6034,10 +7360,10 @@
6034
7360
  "output": 64000
6035
7361
  },
6036
7362
  "cost": {
6037
- "input": 1,
6038
- "output": 5,
6039
- "cache_read": 0.1,
6040
- "cache_write": 1.25
7363
+ "input": 1.1,
7364
+ "output": 5.5,
7365
+ "cache_read": 0.11,
7366
+ "cache_write": 1.375
6041
7367
  }
6042
7368
  },
6043
7369
  "anthropic.claude-fable-5-1": {
@@ -6109,7 +7435,7 @@
6109
7435
  }
6110
7436
  ],
6111
7437
  "tool_call": true,
6112
- "structured_output": true,
7438
+ "structured_output": false,
6113
7439
  "temperature": false,
6114
7440
  "knowledge": "2026-01-31",
6115
7441
  "release_date": "2026-06-30",
@@ -6174,6 +7500,9 @@
6174
7500
  "attachment": true,
6175
7501
  "reasoning": true,
6176
7502
  "reasoning_options": [
7503
+ {
7504
+ "type": "toggle"
7505
+ },
6177
7506
  {
6178
7507
  "type": "budget_tokens",
6179
7508
  "min": 1024
@@ -6201,10 +7530,10 @@
6201
7530
  "output": 64000
6202
7531
  },
6203
7532
  "cost": {
6204
- "input": 3,
6205
- "output": 15,
6206
- "cache_read": 0.3,
6207
- "cache_write": 3.75
7533
+ "input": 3.3,
7534
+ "output": 16.5,
7535
+ "cache_read": 0.33,
7536
+ "cache_write": 4.125
6208
7537
  }
6209
7538
  },
6210
7539
  "global.openai.gpt-6-astra": {
@@ -6235,8 +7564,7 @@
6235
7564
  "modalities": {
6236
7565
  "input": [
6237
7566
  "text",
6238
- "image",
6239
- "pdf"
7567
+ "image"
6240
7568
  ],
6241
7569
  "output": [
6242
7570
  "text"
@@ -6273,6 +7601,73 @@
6273
7601
  }
6274
7602
  }
6275
7603
  },
7604
+ "in.openai.gpt-5.6-luna": {
7605
+ "id": "in.openai.gpt-5.6-luna",
7606
+ "name": "GPT-5.6 Luna (India)",
7607
+ "description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
7608
+ "family": "gpt-luna",
7609
+ "attachment": true,
7610
+ "reasoning": true,
7611
+ "reasoning_options": [
7612
+ {
7613
+ "type": "effort",
7614
+ "values": [
7615
+ "none",
7616
+ "low",
7617
+ "medium",
7618
+ "high",
7619
+ "xhigh",
7620
+ "max"
7621
+ ]
7622
+ }
7623
+ ],
7624
+ "tool_call": true,
7625
+ "structured_output": true,
7626
+ "temperature": false,
7627
+ "knowledge": "2026-02-16",
7628
+ "release_date": "2026-07-09",
7629
+ "last_updated": "2026-07-09",
7630
+ "modalities": {
7631
+ "input": [
7632
+ "text",
7633
+ "image",
7634
+ "pdf"
7635
+ ],
7636
+ "output": [
7637
+ "text"
7638
+ ]
7639
+ },
7640
+ "open_weights": false,
7641
+ "limit": {
7642
+ "context": 1050000,
7643
+ "input": 922000,
7644
+ "output": 128000
7645
+ },
7646
+ "cost": {
7647
+ "input": 0.22,
7648
+ "output": 1.32,
7649
+ "cache_read": 0.022,
7650
+ "cache_write": 0.275,
7651
+ "tiers": [
7652
+ {
7653
+ "input": 0.44,
7654
+ "output": 1.98,
7655
+ "cache_read": 0.044,
7656
+ "cache_write": 0.55,
7657
+ "tier": {
7658
+ "type": "context",
7659
+ "size": 272000
7660
+ }
7661
+ }
7662
+ ],
7663
+ "context_over_200k": {
7664
+ "input": 0.44,
7665
+ "output": 1.98,
7666
+ "cache_read": 0.044,
7667
+ "cache_write": 0.55
7668
+ }
7669
+ }
7670
+ },
6276
7671
  "eu.anthropic.claude-sonnet-4-5-20250929-v1:0": {
6277
7672
  "id": "eu.anthropic.claude-sonnet-4-5-20250929-v1:0",
6278
7673
  "name": "Claude Sonnet 4.5 (EU)",
@@ -6281,6 +7676,9 @@
6281
7676
  "attachment": true,
6282
7677
  "reasoning": true,
6283
7678
  "reasoning_options": [
7679
+ {
7680
+ "type": "toggle"
7681
+ },
6284
7682
  {
6285
7683
  "type": "budget_tokens",
6286
7684
  "min": 1024
@@ -6316,17 +7714,22 @@
6316
7714
  },
6317
7715
  "deepseek.v3.2": {
6318
7716
  "id": "deepseek.v3.2",
6319
- "name": "DeepSeek-V3.2",
6320
- "description": "DeepSeek chat model for instruction following, coding, and analysis",
7717
+ "name": "DeepSeek V3.2",
7718
+ "description": "Hybrid-reasoning DeepSeek model with thinking and non-thinking modes, sparse attention, and tool-use",
6321
7719
  "family": "deepseek",
6322
7720
  "attachment": false,
6323
7721
  "reasoning": true,
6324
- "reasoning_options": [],
7722
+ "reasoning_options": [
7723
+ {
7724
+ "type": "toggle"
7725
+ }
7726
+ ],
6325
7727
  "tool_call": true,
7728
+ "interleaved": true,
6326
7729
  "structured_output": true,
6327
7730
  "temperature": true,
6328
7731
  "knowledge": "2024-07",
6329
- "release_date": "2026-02-06",
7732
+ "release_date": "2025-12-01",
6330
7733
  "last_updated": "2026-02-06",
6331
7734
  "modalities": {
6332
7735
  "input": [