@just-every/ensemble 0.2.260 → 0.2.262

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.
@@ -1 +1 @@
1
- {"version":3,"file":"model_data.d.ts","sourceRoot":"","sources":["../../data/model_data.ts"],"names":[],"mappings":"AAYA,OAAO,EACH,YAAY,EACZ,eAAe,EACf,WAAW,EACX,cAAc,EACd,aAAa,EACb,SAAS,EACT,aAAa,EACb,UAAU,EACV,UAAU,EACV,UAAU,EACb,MAAM,mBAAmB,CAAC;AAM3B,YAAY,EACR,YAAY,EACZ,eAAe,EACf,WAAW,EACX,cAAc,EACd,aAAa,EACb,SAAS,EACT,aAAa,EACb,UAAU,EACV,UAAU,EACV,UAAU,GACb,CAAC;AAOF,eAAO,MAAM,aAAa;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;CA0MzB,CAAC;AAGF,eAAO,MAAM,cAAc,EAAE,UAAU,EA+lHtC,CAAC;AAQF,wBAAgB,SAAS,CAAC,OAAO,EAAE,MAAM,GAAG,UAAU,GAAG,SAAS,CAuBjE"}
1
+ {"version":3,"file":"model_data.d.ts","sourceRoot":"","sources":["../../data/model_data.ts"],"names":[],"mappings":"AAYA,OAAO,EACH,YAAY,EACZ,eAAe,EACf,WAAW,EACX,cAAc,EACd,aAAa,EACb,SAAS,EACT,aAAa,EACb,UAAU,EACV,UAAU,EACV,UAAU,EACb,MAAM,mBAAmB,CAAC;AAM3B,YAAY,EACR,YAAY,EACZ,eAAe,EACf,WAAW,EACX,cAAc,EACd,aAAa,EACb,SAAS,EACT,aAAa,EACb,UAAU,EACV,UAAU,EACV,UAAU,GACb,CAAC;AAOF,eAAO,MAAM,aAAa;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;CA0MzB,CAAC;AAGF,eAAO,MAAM,cAAc,EAAE,UAAU,EA6xHtC,CAAC;AAQF,wBAAgB,SAAS,CAAC,OAAO,EAAE,MAAM,GAAG,UAAU,GAAG,SAAS,CAuBjE"}
@@ -4,8 +4,8 @@ export const MODEL_CLASSES = {
4
4
  models: [
5
5
  'gpt-5.5',
6
6
  'gemini-3-flash-preview',
7
- 'claude-sonnet-4-6',
8
- 'grok-4.3',
7
+ 'claude-sonnet-5',
8
+ 'grok-4.5',
9
9
  ],
10
10
  random: true,
11
11
  },
@@ -22,8 +22,8 @@ export const MODEL_CLASSES = {
22
22
  models: [
23
23
  'gpt-5.5',
24
24
  'gemini-2.5-pro',
25
- 'claude-sonnet-4-6',
26
- 'grok-4.3',
25
+ 'claude-sonnet-5',
26
+ 'grok-4.5',
27
27
  ],
28
28
  random: true,
29
29
  },
@@ -32,7 +32,7 @@ export const MODEL_CLASSES = {
32
32
  'gpt-5.5-pro',
33
33
  'gemini-3.1-pro-preview',
34
34
  'claude-fable-5',
35
- 'grok-4.3',
35
+ 'grok-4.5',
36
36
  ],
37
37
  random: true,
38
38
  },
@@ -40,7 +40,7 @@ export const MODEL_CLASSES = {
40
40
  models: [
41
41
  'gpt-5.4-mini',
42
42
  'gemini-3-flash-preview',
43
- 'claude-sonnet-4-6',
43
+ 'claude-sonnet-5',
44
44
  'grok-3-mini',
45
45
  ],
46
46
  random: true,
@@ -49,8 +49,8 @@ export const MODEL_CLASSES = {
49
49
  models: [
50
50
  'gpt-5.5',
51
51
  'gemini-3.1-pro-preview',
52
- 'claude-sonnet-4-6',
53
- 'grok-4.3',
52
+ 'claude-sonnet-5',
53
+ 'grok-4.5',
54
54
  ],
55
55
  random: true,
56
56
  },
@@ -59,7 +59,7 @@ export const MODEL_CLASSES = {
59
59
  'gpt-5.5',
60
60
  'gemini-3.1-pro-preview',
61
61
  'claude-fable-5',
62
- 'grok-4.3',
62
+ 'grok-4.5',
63
63
  ],
64
64
  random: true,
65
65
  },
@@ -68,7 +68,7 @@ export const MODEL_CLASSES = {
68
68
  'gpt-5.3-codex',
69
69
  'gemini-3.1-pro-preview',
70
70
  'claude-fable-5',
71
- 'grok-4.3',
71
+ 'grok-4.5',
72
72
  'qwen3-coder',
73
73
  ],
74
74
  random: true,
@@ -77,8 +77,8 @@ export const MODEL_CLASSES = {
77
77
  models: [
78
78
  'gpt-5.5',
79
79
  'gemini-3-flash-preview',
80
- 'claude-sonnet-4-6',
81
- 'grok-4.3',
80
+ 'claude-sonnet-5',
81
+ 'grok-4.5',
82
82
  ],
83
83
  random: true,
84
84
  },
@@ -96,7 +96,7 @@ export const MODEL_CLASSES = {
96
96
  'gpt-5.5',
97
97
  'gemini-3.1-pro-preview',
98
98
  'claude-fable-5',
99
- 'grok-4.3',
99
+ 'grok-4.5',
100
100
  ],
101
101
  random: true,
102
102
  },
@@ -153,7 +153,7 @@ export const MODEL_CLASSES = {
153
153
  models: [
154
154
  'tts-1-hd',
155
155
  'eleven_multilingual_v2',
156
- 'gemini-2.5-pro-preview-tts',
156
+ 'gemini-3.1-flash-tts-preview',
157
157
  ],
158
158
  description: 'Text-to-Speech models for voice generation',
159
159
  },
@@ -162,7 +162,7 @@ export const MODEL_CLASSES = {
162
162
  'gpt-realtime-whisper',
163
163
  'gpt-4o-transcribe',
164
164
  'u3-rt-pro',
165
- 'gemini-2.5-flash-native-audio-preview-12-2025',
165
+ 'gemini-3.1-flash-live-preview',
166
166
  ],
167
167
  description: 'Speech-to-Text models for audio transcription with real-time streaming',
168
168
  },
@@ -494,6 +494,27 @@ export const MODEL_REGISTRY = [
494
494
  },
495
495
  description: 'Qwen3.5-397B-A17B is a vision-language 397B MoE model with a native 262k-context window, multimodal input support, and strong reasoning.',
496
496
  },
497
+ {
498
+ id: 'grok-4.5',
499
+ aliases: ['grok-4.5-latest', 'grok-build-latest'],
500
+ provider: 'xai',
501
+ cost: {
502
+ input_per_million: 2.0,
503
+ output_per_million: 6.0,
504
+ cached_input_per_million: 0.5,
505
+ },
506
+ features: {
507
+ context_length: 500_000,
508
+ input_modality: ['text', 'image'],
509
+ output_modality: ['text'],
510
+ tool_use: true,
511
+ streaming: true,
512
+ json_output: true,
513
+ reasoning_output: true,
514
+ },
515
+ class: 'reasoning',
516
+ description: "Grok 4.5, xAI's flagship model for code and general chat with 500k context, multimodal input, tools, structured output, and configurable low/medium/high reasoning.",
517
+ },
497
518
  {
498
519
  id: 'grok-4.3',
499
520
  provider: 'xai',
@@ -514,7 +535,7 @@ export const MODEL_REGISTRY = [
514
535
  },
515
536
  {
516
537
  id: 'grok-build-0.1',
517
- aliases: ['grok-build', 'grok-build-latest'],
538
+ aliases: ['grok-build'],
518
539
  provider: 'xai',
519
540
  cost: {
520
541
  input_per_million: 1.0,
@@ -1798,7 +1819,7 @@ export const MODEL_REGISTRY = [
1798
1819
  },
1799
1820
  {
1800
1821
  id: 'claude-sonnet-4-6',
1801
- aliases: ['claude-sonnet-4-6', 'claude-sonnet-4.6', 'claude-sonnet-latest', 'sonnet-4-6', 'sonnet-4.6'],
1822
+ aliases: ['claude-sonnet-4-6', 'claude-sonnet-4.6', 'sonnet-4-6', 'sonnet-4.6'],
1802
1823
  provider: 'anthropic',
1803
1824
  cost: {
1804
1825
  input_per_million: 3.0,
@@ -1819,6 +1840,34 @@ export const MODEL_REGISTRY = [
1819
1840
  score: 94,
1820
1841
  description: 'Claude Sonnet 4.6 with updated reasoning and multimodal capabilities.',
1821
1842
  },
1843
+ {
1844
+ id: 'claude-sonnet-5',
1845
+ aliases: ['claude-sonnet-latest', 'sonnet-5', 'sonnet-5-latest'],
1846
+ provider: 'anthropic',
1847
+ cost: {
1848
+ input_per_million: 2.0,
1849
+ output_per_million: 10.0,
1850
+ cached_input_per_million: 0.2,
1851
+ },
1852
+ features: {
1853
+ context_length: 1_000_000,
1854
+ input_modality: ['text', 'image'],
1855
+ output_modality: ['text'],
1856
+ tool_use: true,
1857
+ streaming: true,
1858
+ json_output: true,
1859
+ reasoning_output: true,
1860
+ max_output_tokens: 128000,
1861
+ },
1862
+ class: 'reasoning',
1863
+ score: 96,
1864
+ scores: {
1865
+ monologue: 95,
1866
+ code: 96,
1867
+ reasoning: 95,
1868
+ },
1869
+ description: 'Claude Sonnet 5 - latest Sonnet model with 1M context, 128K output, vision support, and implicit adaptive thinking controls.',
1870
+ },
1822
1871
  {
1823
1872
  id: 'claude-sonnet-4-5-20250514-long-context',
1824
1873
  aliases: ['claude-sonnet-4-5-long', 'claude-sonnet-4.5-long'],
@@ -2637,6 +2686,22 @@ export const MODEL_REGISTRY = [
2637
2686
  class: 'voice',
2638
2687
  description: "Gemini's advanced text-to-speech model with superior voice quality, expression control, and multi-speaker support for creating dynamic conversations.",
2639
2688
  },
2689
+ {
2690
+ id: 'gemini-3.1-flash-tts-preview',
2691
+ provider: 'google',
2692
+ cost: {
2693
+ input_per_million: 10.0,
2694
+ output_per_million: 0,
2695
+ },
2696
+ features: {
2697
+ input_modality: ['text'],
2698
+ output_modality: ['audio'],
2699
+ streaming: true,
2700
+ context_length: 32000,
2701
+ },
2702
+ class: 'voice',
2703
+ description: 'Gemini 3.1 Flash TTS preview for low-latency speech generation with steerable prompts and expressive audio tags.',
2704
+ },
2640
2705
  {
2641
2706
  id: 'gemini-2.5-flash-native-audio-preview',
2642
2707
  aliases: ['gemini-2.5-flash-native-audio-preview-12-2025'],
@@ -2654,6 +2719,21 @@ export const MODEL_REGISTRY = [
2654
2719
  class: 'voice',
2655
2720
  description: "Gemini's native audio preview model providing low-latency text-to-speech with built-in audio processing.",
2656
2721
  },
2722
+ {
2723
+ id: 'gemini-omni-flash',
2724
+ provider: 'google',
2725
+ cost: {
2726
+ input_per_million: 0,
2727
+ output_per_million: 0,
2728
+ },
2729
+ features: {
2730
+ input_modality: ['text', 'image', 'audio', 'video'],
2731
+ output_modality: ['text', 'image', 'audio'],
2732
+ streaming: true,
2733
+ },
2734
+ class: 'standard',
2735
+ description: 'Gemini Omni Flash preview for multimodal, conversational generation and editing across text, image, audio, and video inputs.',
2736
+ },
2657
2737
  {
2658
2738
  id: 'perplexity/sonar',
2659
2739
  provider: 'openrouter',
@@ -2996,6 +3076,29 @@ export const MODEL_REGISTRY = [
2996
3076
  class: 'transcription',
2997
3077
  description: 'Gemini Live API for real-time multimodal interaction with modality-specific pricing',
2998
3078
  },
3079
+ {
3080
+ id: 'gemini-3.1-flash-live-preview',
3081
+ provider: 'google',
3082
+ cost: {
3083
+ input_per_million: {
3084
+ text: 0.5,
3085
+ audio: 3.0,
3086
+ video: 3.0,
3087
+ },
3088
+ output_per_million: {
3089
+ text: 2.0,
3090
+ audio: 12.0,
3091
+ },
3092
+ },
3093
+ features: {
3094
+ context_length: 32000,
3095
+ input_modality: ['text', 'audio', 'video'],
3096
+ output_modality: ['text', 'audio'],
3097
+ streaming: true,
3098
+ },
3099
+ class: 'transcription',
3100
+ description: 'Gemini 3.1 Flash Live preview for high-quality, low-latency audio-to-audio realtime dialogue and voice-first applications.',
3101
+ },
2999
3102
  {
3000
3103
  id: 'gemini-2.0-flash-live-001',
3001
3104
  provider: 'google',
@@ -3173,8 +3276,8 @@ export const MODEL_REGISTRY = [
3173
3276
  provider: 'openrouter',
3174
3277
  openrouter_id: 'openai/gpt-oss-120b',
3175
3278
  cost: {
3176
- input_per_million: 0.039,
3177
- output_per_million: 0.18,
3279
+ input_per_million: 0.03,
3280
+ output_per_million: 0.15,
3178
3281
  },
3179
3282
  features: {
3180
3283
  context_length: 131072,
@@ -3233,6 +3336,29 @@ export const MODEL_REGISTRY = [
3233
3336
  class: 'code',
3234
3337
  description: 'Qwen3.7 Max via OpenRouter. Flagship Qwen 3.7 model for agent-centric coding, productivity, and long-horizon autonomous execution.',
3235
3338
  },
3339
+ {
3340
+ id: 'qwen/qwen3.7-plus',
3341
+ aliases: ['Qwen3.7 Plus', 'qwen3.7-plus', 'qwen-3.7-plus'],
3342
+ provider: 'openrouter',
3343
+ openrouter_id: 'qwen/qwen3.7-plus',
3344
+ cost: {
3345
+ input_per_million: 0.32,
3346
+ cached_input_per_million: 0.064,
3347
+ output_per_million: 1.28,
3348
+ },
3349
+ features: {
3350
+ context_length: 1000000,
3351
+ input_modality: ['text', 'image'],
3352
+ output_modality: ['text'],
3353
+ tool_use: true,
3354
+ streaming: true,
3355
+ json_output: true,
3356
+ max_output_tokens: 65536,
3357
+ reasoning_output: true,
3358
+ },
3359
+ class: 'reasoning',
3360
+ description: 'Qwen3.7 Plus via OpenRouter. Current multimodal Qwen 3.7 model with 1M context for agentic workflows and visual reasoning.',
3361
+ },
3236
3362
  {
3237
3363
  id: 'qwen/qwen3.6-plus',
3238
3364
  aliases: ['Qwen 3.6', 'Qwen3.6', 'qwen3.6', 'qwen-3.6', 'qwen3.6-plus', 'qwen-3.6-plus'],
@@ -3293,7 +3419,7 @@ export const MODEL_REGISTRY = [
3293
3419
  tool_use: true,
3294
3420
  streaming: true,
3295
3421
  json_output: true,
3296
- max_output_tokens: 262140,
3422
+ max_output_tokens: 262144,
3297
3423
  reasoning_output: true,
3298
3424
  },
3299
3425
  class: 'reasoning',
@@ -3328,8 +3454,9 @@ export const MODEL_REGISTRY = [
3328
3454
  provider: 'openrouter',
3329
3455
  openrouter_id: 'qwen/qwen3.6-27b',
3330
3456
  cost: {
3331
- input_per_million: 0.29,
3332
- output_per_million: 3.2,
3457
+ input_per_million: 0.285,
3458
+ cached_input_per_million: 0.15,
3459
+ output_per_million: 2.4,
3333
3460
  },
3334
3461
  features: {
3335
3462
  context_length: 262144,
@@ -3392,13 +3519,13 @@ export const MODEL_REGISTRY = [
3392
3519
  },
3393
3520
  {
3394
3521
  id: 'z-ai/glm-5.1',
3395
- aliases: ['GLM-5.1', 'glm-5.1', 'glm-5', 'z-ai/glm-5'],
3522
+ aliases: ['GLM-5.1', 'glm-5.1'],
3396
3523
  provider: 'openrouter',
3397
3524
  openrouter_id: 'z-ai/glm-5.1',
3398
3525
  cost: {
3399
- input_per_million: 0.98,
3400
- output_per_million: 3.08,
3401
- cached_input_per_million: 0.182,
3526
+ input_per_million: 0.966,
3527
+ output_per_million: 3.036,
3528
+ cached_input_per_million: 0.1794,
3402
3529
  },
3403
3530
  features: {
3404
3531
  context_length: 202752,
@@ -3412,6 +3539,29 @@ export const MODEL_REGISTRY = [
3412
3539
  class: 'reasoning',
3413
3540
  description: 'GLM-5.1 via OpenRouter. Z.ai flagship text model for long-horizon autonomous coding and sustained agent workflows.',
3414
3541
  },
3542
+ {
3543
+ id: 'z-ai/glm-5.2',
3544
+ aliases: ['GLM-5.2', 'glm-5.2', 'glm-5'],
3545
+ provider: 'openrouter',
3546
+ openrouter_id: 'z-ai/glm-5.2',
3547
+ cost: {
3548
+ input_per_million: 0.93,
3549
+ output_per_million: 3.0,
3550
+ cached_input_per_million: 0.18,
3551
+ },
3552
+ features: {
3553
+ context_length: 1048576,
3554
+ input_modality: ['text'],
3555
+ output_modality: ['text'],
3556
+ tool_use: true,
3557
+ streaming: true,
3558
+ json_output: true,
3559
+ max_output_tokens: 32768,
3560
+ reasoning_output: true,
3561
+ },
3562
+ class: 'reasoning',
3563
+ description: 'GLM-5.2 via OpenRouter. Z.ai 1M-context text model for long-horizon autonomous coding and sustained agent workflows.',
3564
+ },
3415
3565
  {
3416
3566
  id: 'deepseek/deepseek-v4-pro',
3417
3567
  aliases: ['DeepSeek-V4', 'DeepSeek-V4-Pro', 'deepseek-v4', 'deepseek-v4-pro'],
@@ -3441,9 +3591,9 @@ export const MODEL_REGISTRY = [
3441
3591
  provider: 'openrouter',
3442
3592
  openrouter_id: 'deepseek/deepseek-v4-flash',
3443
3593
  cost: {
3444
- input_per_million: 0.0983,
3445
- cached_input_per_million: 0.0197,
3446
- output_per_million: 0.1966,
3594
+ input_per_million: 0.09,
3595
+ cached_input_per_million: 0.018,
3596
+ output_per_million: 0.18,
3447
3597
  },
3448
3598
  features: {
3449
3599
  context_length: 1048576,
@@ -3464,8 +3614,8 @@ export const MODEL_REGISTRY = [
3464
3614
  provider: 'openrouter',
3465
3615
  openrouter_id: 'xiaomi/mimo-v2.5',
3466
3616
  cost: {
3467
- input_per_million: 0.14,
3468
- cached_input_per_million: 0.0028,
3617
+ input_per_million: 0.105,
3618
+ cached_input_per_million: 0.028,
3469
3619
  output_per_million: 0.28,
3470
3620
  },
3471
3621
  features: {
@@ -3529,14 +3679,14 @@ export const MODEL_REGISTRY = [
3529
3679
  description: 'MiniMax M3 via OpenRouter. Multimodal 1M-context foundation model for long-horizon agentic work, coding, and video-aware reasoning.',
3530
3680
  },
3531
3681
  {
3532
- id: 'tencent/hy3-preview',
3533
- aliases: ['Tencent Hy3 Preview', 'Hy3 Preview', 'Hy3-preview', 'hy3-preview'],
3682
+ id: 'tencent/hy3',
3683
+ aliases: ['Tencent Hy3', 'Hy3', 'hy3', 'Tencent Hy3 Preview', 'Hy3 Preview', 'Hy3-preview', 'hy3-preview'],
3534
3684
  provider: 'openrouter',
3535
- openrouter_id: 'tencent/hy3-preview',
3685
+ openrouter_id: 'tencent/hy3',
3536
3686
  cost: {
3537
- input_per_million: 0.063,
3538
- cached_input_per_million: 0.021,
3539
- output_per_million: 0.21,
3687
+ input_per_million: 0.2,
3688
+ cached_input_per_million: 0.5,
3689
+ output_per_million: 0.8,
3540
3690
  },
3541
3691
  features: {
3542
3692
  context_length: 262144,
@@ -3545,19 +3695,20 @@ export const MODEL_REGISTRY = [
3545
3695
  tool_use: true,
3546
3696
  streaming: true,
3547
3697
  json_output: false,
3698
+ max_output_tokens: 131072,
3548
3699
  reasoning_output: true,
3549
3700
  },
3550
3701
  class: 'reasoning',
3551
- description: 'Tencent Hy3 preview via OpenRouter. High-efficiency MoE model for agentic workflows with configurable disabled, low, and high reasoning modes.',
3702
+ description: 'Tencent Hy3 via OpenRouter. High-efficiency text model for agentic workflows with configurable disabled, low, and high reasoning modes.',
3552
3703
  },
3553
3704
  {
3554
3705
  id: 'moonshotai/kimi-k2.5',
3555
3706
  aliases: ['kimi-k2.5', 'kimi-k2-5'],
3556
3707
  provider: 'openrouter',
3557
3708
  cost: {
3558
- input_per_million: 0.4,
3559
- output_per_million: 1.9,
3560
- cached_input_per_million: 0.09,
3709
+ input_per_million: 0.375,
3710
+ output_per_million: 2.025,
3711
+ cached_input_per_million: 0.203,
3561
3712
  },
3562
3713
  features: {
3563
3714
  context_length: 262144,
@@ -3578,9 +3729,9 @@ export const MODEL_REGISTRY = [
3578
3729
  provider: 'openrouter',
3579
3730
  openrouter_id: 'moonshotai/kimi-k2.6',
3580
3731
  cost: {
3581
- input_per_million: 0.684,
3582
- output_per_million: 3.42,
3583
- cached_input_per_million: 0.144,
3732
+ input_per_million: 0.66,
3733
+ output_per_million: 3.41,
3734
+ cached_input_per_million: 0.14,
3584
3735
  },
3585
3736
  features: {
3586
3737
  context_length: 262144,
@@ -3595,6 +3746,29 @@ export const MODEL_REGISTRY = [
3595
3746
  class: 'reasoning',
3596
3747
  description: 'MoonshotAI Kimi K2.6 via OpenRouter. Multimodal long-horizon coding and multi-agent orchestration model with 262K context.',
3597
3748
  },
3749
+ {
3750
+ id: 'moonshotai/kimi-k2.7-code',
3751
+ aliases: ['Kimi K2.7 Code', 'kimi-k2.7-code', 'kimi-k2-7-code', 'kimi-k2.7', 'kimi-k2-7'],
3752
+ provider: 'openrouter',
3753
+ openrouter_id: 'moonshotai/kimi-k2.7-code',
3754
+ cost: {
3755
+ input_per_million: 0.74,
3756
+ output_per_million: 3.5,
3757
+ cached_input_per_million: 0.15,
3758
+ },
3759
+ features: {
3760
+ context_length: 262144,
3761
+ input_modality: ['text', 'image'],
3762
+ output_modality: ['text'],
3763
+ tool_use: true,
3764
+ streaming: true,
3765
+ json_output: true,
3766
+ max_output_tokens: 16384,
3767
+ reasoning_output: true,
3768
+ },
3769
+ class: 'code',
3770
+ description: 'MoonshotAI Kimi K2.7 Code via OpenRouter. Coding-focused long-context model for software engineering and agentic coding workflows.',
3771
+ },
3598
3772
  ];
3599
3773
  export function findModel(modelId) {
3600
3774
  const externalModel = getExternalModel(modelId);