@just-every/ensemble 0.2.260 → 0.2.261

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,EAswHtC,CAAC;AAQF,wBAAgB,SAAS,CAAC,OAAO,EAAE,MAAM,GAAG,UAAU,GAAG,SAAS,CAuBjE"}
@@ -4,7 +4,7 @@ export const MODEL_CLASSES = {
4
4
  models: [
5
5
  'gpt-5.5',
6
6
  'gemini-3-flash-preview',
7
- 'claude-sonnet-4-6',
7
+ 'claude-sonnet-5',
8
8
  'grok-4.3',
9
9
  ],
10
10
  random: true,
@@ -22,7 +22,7 @@ export const MODEL_CLASSES = {
22
22
  models: [
23
23
  'gpt-5.5',
24
24
  'gemini-2.5-pro',
25
- 'claude-sonnet-4-6',
25
+ 'claude-sonnet-5',
26
26
  'grok-4.3',
27
27
  ],
28
28
  random: true,
@@ -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,
@@ -49,7 +49,7 @@ export const MODEL_CLASSES = {
49
49
  models: [
50
50
  'gpt-5.5',
51
51
  'gemini-3.1-pro-preview',
52
- 'claude-sonnet-4-6',
52
+ 'claude-sonnet-5',
53
53
  'grok-4.3',
54
54
  ],
55
55
  random: true,
@@ -77,7 +77,7 @@ export const MODEL_CLASSES = {
77
77
  models: [
78
78
  'gpt-5.5',
79
79
  'gemini-3-flash-preview',
80
- 'claude-sonnet-4-6',
80
+ 'claude-sonnet-5',
81
81
  'grok-4.3',
82
82
  ],
83
83
  random: true,
@@ -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
  },
@@ -1798,7 +1798,7 @@ export const MODEL_REGISTRY = [
1798
1798
  },
1799
1799
  {
1800
1800
  id: 'claude-sonnet-4-6',
1801
- aliases: ['claude-sonnet-4-6', 'claude-sonnet-4.6', 'claude-sonnet-latest', 'sonnet-4-6', 'sonnet-4.6'],
1801
+ aliases: ['claude-sonnet-4-6', 'claude-sonnet-4.6', 'sonnet-4-6', 'sonnet-4.6'],
1802
1802
  provider: 'anthropic',
1803
1803
  cost: {
1804
1804
  input_per_million: 3.0,
@@ -1819,6 +1819,34 @@ export const MODEL_REGISTRY = [
1819
1819
  score: 94,
1820
1820
  description: 'Claude Sonnet 4.6 with updated reasoning and multimodal capabilities.',
1821
1821
  },
1822
+ {
1823
+ id: 'claude-sonnet-5',
1824
+ aliases: ['claude-sonnet-latest', 'sonnet-5', 'sonnet-5-latest'],
1825
+ provider: 'anthropic',
1826
+ cost: {
1827
+ input_per_million: 2.0,
1828
+ output_per_million: 10.0,
1829
+ cached_input_per_million: 0.2,
1830
+ },
1831
+ features: {
1832
+ context_length: 1_000_000,
1833
+ input_modality: ['text', 'image'],
1834
+ output_modality: ['text'],
1835
+ tool_use: true,
1836
+ streaming: true,
1837
+ json_output: true,
1838
+ reasoning_output: true,
1839
+ max_output_tokens: 128000,
1840
+ },
1841
+ class: 'reasoning',
1842
+ score: 96,
1843
+ scores: {
1844
+ monologue: 95,
1845
+ code: 96,
1846
+ reasoning: 95,
1847
+ },
1848
+ description: 'Claude Sonnet 5 - latest Sonnet model with 1M context, 128K output, vision support, and implicit adaptive thinking controls.',
1849
+ },
1822
1850
  {
1823
1851
  id: 'claude-sonnet-4-5-20250514-long-context',
1824
1852
  aliases: ['claude-sonnet-4-5-long', 'claude-sonnet-4.5-long'],
@@ -2637,6 +2665,22 @@ export const MODEL_REGISTRY = [
2637
2665
  class: 'voice',
2638
2666
  description: "Gemini's advanced text-to-speech model with superior voice quality, expression control, and multi-speaker support for creating dynamic conversations.",
2639
2667
  },
2668
+ {
2669
+ id: 'gemini-3.1-flash-tts-preview',
2670
+ provider: 'google',
2671
+ cost: {
2672
+ input_per_million: 10.0,
2673
+ output_per_million: 0,
2674
+ },
2675
+ features: {
2676
+ input_modality: ['text'],
2677
+ output_modality: ['audio'],
2678
+ streaming: true,
2679
+ context_length: 32000,
2680
+ },
2681
+ class: 'voice',
2682
+ description: 'Gemini 3.1 Flash TTS preview for low-latency speech generation with steerable prompts and expressive audio tags.',
2683
+ },
2640
2684
  {
2641
2685
  id: 'gemini-2.5-flash-native-audio-preview',
2642
2686
  aliases: ['gemini-2.5-flash-native-audio-preview-12-2025'],
@@ -2654,6 +2698,21 @@ export const MODEL_REGISTRY = [
2654
2698
  class: 'voice',
2655
2699
  description: "Gemini's native audio preview model providing low-latency text-to-speech with built-in audio processing.",
2656
2700
  },
2701
+ {
2702
+ id: 'gemini-omni-flash',
2703
+ provider: 'google',
2704
+ cost: {
2705
+ input_per_million: 0,
2706
+ output_per_million: 0,
2707
+ },
2708
+ features: {
2709
+ input_modality: ['text', 'image', 'audio', 'video'],
2710
+ output_modality: ['text', 'image', 'audio'],
2711
+ streaming: true,
2712
+ },
2713
+ class: 'standard',
2714
+ description: 'Gemini Omni Flash preview for multimodal, conversational generation and editing across text, image, audio, and video inputs.',
2715
+ },
2657
2716
  {
2658
2717
  id: 'perplexity/sonar',
2659
2718
  provider: 'openrouter',
@@ -2996,6 +3055,29 @@ export const MODEL_REGISTRY = [
2996
3055
  class: 'transcription',
2997
3056
  description: 'Gemini Live API for real-time multimodal interaction with modality-specific pricing',
2998
3057
  },
3058
+ {
3059
+ id: 'gemini-3.1-flash-live-preview',
3060
+ provider: 'google',
3061
+ cost: {
3062
+ input_per_million: {
3063
+ text: 0.5,
3064
+ audio: 3.0,
3065
+ video: 3.0,
3066
+ },
3067
+ output_per_million: {
3068
+ text: 2.0,
3069
+ audio: 12.0,
3070
+ },
3071
+ },
3072
+ features: {
3073
+ context_length: 32000,
3074
+ input_modality: ['text', 'audio', 'video'],
3075
+ output_modality: ['text', 'audio'],
3076
+ streaming: true,
3077
+ },
3078
+ class: 'transcription',
3079
+ description: 'Gemini 3.1 Flash Live preview for high-quality, low-latency audio-to-audio realtime dialogue and voice-first applications.',
3080
+ },
2999
3081
  {
3000
3082
  id: 'gemini-2.0-flash-live-001',
3001
3083
  provider: 'google',
@@ -3173,8 +3255,8 @@ export const MODEL_REGISTRY = [
3173
3255
  provider: 'openrouter',
3174
3256
  openrouter_id: 'openai/gpt-oss-120b',
3175
3257
  cost: {
3176
- input_per_million: 0.039,
3177
- output_per_million: 0.18,
3258
+ input_per_million: 0.03,
3259
+ output_per_million: 0.15,
3178
3260
  },
3179
3261
  features: {
3180
3262
  context_length: 131072,
@@ -3233,6 +3315,29 @@ export const MODEL_REGISTRY = [
3233
3315
  class: 'code',
3234
3316
  description: 'Qwen3.7 Max via OpenRouter. Flagship Qwen 3.7 model for agent-centric coding, productivity, and long-horizon autonomous execution.',
3235
3317
  },
3318
+ {
3319
+ id: 'qwen/qwen3.7-plus',
3320
+ aliases: ['Qwen3.7 Plus', 'qwen3.7-plus', 'qwen-3.7-plus'],
3321
+ provider: 'openrouter',
3322
+ openrouter_id: 'qwen/qwen3.7-plus',
3323
+ cost: {
3324
+ input_per_million: 0.32,
3325
+ cached_input_per_million: 0.064,
3326
+ output_per_million: 1.28,
3327
+ },
3328
+ features: {
3329
+ context_length: 1000000,
3330
+ input_modality: ['text', 'image'],
3331
+ output_modality: ['text'],
3332
+ tool_use: true,
3333
+ streaming: true,
3334
+ json_output: true,
3335
+ max_output_tokens: 65536,
3336
+ reasoning_output: true,
3337
+ },
3338
+ class: 'reasoning',
3339
+ description: 'Qwen3.7 Plus via OpenRouter. Current multimodal Qwen 3.7 model with 1M context for agentic workflows and visual reasoning.',
3340
+ },
3236
3341
  {
3237
3342
  id: 'qwen/qwen3.6-plus',
3238
3343
  aliases: ['Qwen 3.6', 'Qwen3.6', 'qwen3.6', 'qwen-3.6', 'qwen3.6-plus', 'qwen-3.6-plus'],
@@ -3293,7 +3398,7 @@ export const MODEL_REGISTRY = [
3293
3398
  tool_use: true,
3294
3399
  streaming: true,
3295
3400
  json_output: true,
3296
- max_output_tokens: 262140,
3401
+ max_output_tokens: 262144,
3297
3402
  reasoning_output: true,
3298
3403
  },
3299
3404
  class: 'reasoning',
@@ -3328,8 +3433,9 @@ export const MODEL_REGISTRY = [
3328
3433
  provider: 'openrouter',
3329
3434
  openrouter_id: 'qwen/qwen3.6-27b',
3330
3435
  cost: {
3331
- input_per_million: 0.29,
3332
- output_per_million: 3.2,
3436
+ input_per_million: 0.285,
3437
+ cached_input_per_million: 0.15,
3438
+ output_per_million: 2.4,
3333
3439
  },
3334
3440
  features: {
3335
3441
  context_length: 262144,
@@ -3392,13 +3498,13 @@ export const MODEL_REGISTRY = [
3392
3498
  },
3393
3499
  {
3394
3500
  id: 'z-ai/glm-5.1',
3395
- aliases: ['GLM-5.1', 'glm-5.1', 'glm-5', 'z-ai/glm-5'],
3501
+ aliases: ['GLM-5.1', 'glm-5.1'],
3396
3502
  provider: 'openrouter',
3397
3503
  openrouter_id: 'z-ai/glm-5.1',
3398
3504
  cost: {
3399
- input_per_million: 0.98,
3400
- output_per_million: 3.08,
3401
- cached_input_per_million: 0.182,
3505
+ input_per_million: 0.966,
3506
+ output_per_million: 3.036,
3507
+ cached_input_per_million: 0.1794,
3402
3508
  },
3403
3509
  features: {
3404
3510
  context_length: 202752,
@@ -3412,6 +3518,29 @@ export const MODEL_REGISTRY = [
3412
3518
  class: 'reasoning',
3413
3519
  description: 'GLM-5.1 via OpenRouter. Z.ai flagship text model for long-horizon autonomous coding and sustained agent workflows.',
3414
3520
  },
3521
+ {
3522
+ id: 'z-ai/glm-5.2',
3523
+ aliases: ['GLM-5.2', 'glm-5.2', 'glm-5'],
3524
+ provider: 'openrouter',
3525
+ openrouter_id: 'z-ai/glm-5.2',
3526
+ cost: {
3527
+ input_per_million: 0.93,
3528
+ output_per_million: 3.0,
3529
+ cached_input_per_million: 0.18,
3530
+ },
3531
+ features: {
3532
+ context_length: 1048576,
3533
+ input_modality: ['text'],
3534
+ output_modality: ['text'],
3535
+ tool_use: true,
3536
+ streaming: true,
3537
+ json_output: true,
3538
+ max_output_tokens: 32768,
3539
+ reasoning_output: true,
3540
+ },
3541
+ class: 'reasoning',
3542
+ description: 'GLM-5.2 via OpenRouter. Z.ai 1M-context text model for long-horizon autonomous coding and sustained agent workflows.',
3543
+ },
3415
3544
  {
3416
3545
  id: 'deepseek/deepseek-v4-pro',
3417
3546
  aliases: ['DeepSeek-V4', 'DeepSeek-V4-Pro', 'deepseek-v4', 'deepseek-v4-pro'],
@@ -3441,9 +3570,9 @@ export const MODEL_REGISTRY = [
3441
3570
  provider: 'openrouter',
3442
3571
  openrouter_id: 'deepseek/deepseek-v4-flash',
3443
3572
  cost: {
3444
- input_per_million: 0.0983,
3445
- cached_input_per_million: 0.0197,
3446
- output_per_million: 0.1966,
3573
+ input_per_million: 0.09,
3574
+ cached_input_per_million: 0.018,
3575
+ output_per_million: 0.18,
3447
3576
  },
3448
3577
  features: {
3449
3578
  context_length: 1048576,
@@ -3464,8 +3593,8 @@ export const MODEL_REGISTRY = [
3464
3593
  provider: 'openrouter',
3465
3594
  openrouter_id: 'xiaomi/mimo-v2.5',
3466
3595
  cost: {
3467
- input_per_million: 0.14,
3468
- cached_input_per_million: 0.0028,
3596
+ input_per_million: 0.105,
3597
+ cached_input_per_million: 0.028,
3469
3598
  output_per_million: 0.28,
3470
3599
  },
3471
3600
  features: {
@@ -3529,14 +3658,14 @@ export const MODEL_REGISTRY = [
3529
3658
  description: 'MiniMax M3 via OpenRouter. Multimodal 1M-context foundation model for long-horizon agentic work, coding, and video-aware reasoning.',
3530
3659
  },
3531
3660
  {
3532
- id: 'tencent/hy3-preview',
3533
- aliases: ['Tencent Hy3 Preview', 'Hy3 Preview', 'Hy3-preview', 'hy3-preview'],
3661
+ id: 'tencent/hy3',
3662
+ aliases: ['Tencent Hy3', 'Hy3', 'hy3', 'Tencent Hy3 Preview', 'Hy3 Preview', 'Hy3-preview', 'hy3-preview'],
3534
3663
  provider: 'openrouter',
3535
- openrouter_id: 'tencent/hy3-preview',
3664
+ openrouter_id: 'tencent/hy3',
3536
3665
  cost: {
3537
- input_per_million: 0.063,
3538
- cached_input_per_million: 0.021,
3539
- output_per_million: 0.21,
3666
+ input_per_million: 0.2,
3667
+ cached_input_per_million: 0.5,
3668
+ output_per_million: 0.8,
3540
3669
  },
3541
3670
  features: {
3542
3671
  context_length: 262144,
@@ -3545,19 +3674,20 @@ export const MODEL_REGISTRY = [
3545
3674
  tool_use: true,
3546
3675
  streaming: true,
3547
3676
  json_output: false,
3677
+ max_output_tokens: 131072,
3548
3678
  reasoning_output: true,
3549
3679
  },
3550
3680
  class: 'reasoning',
3551
- description: 'Tencent Hy3 preview via OpenRouter. High-efficiency MoE model for agentic workflows with configurable disabled, low, and high reasoning modes.',
3681
+ description: 'Tencent Hy3 via OpenRouter. High-efficiency text model for agentic workflows with configurable disabled, low, and high reasoning modes.',
3552
3682
  },
3553
3683
  {
3554
3684
  id: 'moonshotai/kimi-k2.5',
3555
3685
  aliases: ['kimi-k2.5', 'kimi-k2-5'],
3556
3686
  provider: 'openrouter',
3557
3687
  cost: {
3558
- input_per_million: 0.4,
3559
- output_per_million: 1.9,
3560
- cached_input_per_million: 0.09,
3688
+ input_per_million: 0.375,
3689
+ output_per_million: 2.025,
3690
+ cached_input_per_million: 0.203,
3561
3691
  },
3562
3692
  features: {
3563
3693
  context_length: 262144,
@@ -3578,9 +3708,9 @@ export const MODEL_REGISTRY = [
3578
3708
  provider: 'openrouter',
3579
3709
  openrouter_id: 'moonshotai/kimi-k2.6',
3580
3710
  cost: {
3581
- input_per_million: 0.684,
3582
- output_per_million: 3.42,
3583
- cached_input_per_million: 0.144,
3711
+ input_per_million: 0.66,
3712
+ output_per_million: 3.41,
3713
+ cached_input_per_million: 0.14,
3584
3714
  },
3585
3715
  features: {
3586
3716
  context_length: 262144,
@@ -3595,6 +3725,29 @@ export const MODEL_REGISTRY = [
3595
3725
  class: 'reasoning',
3596
3726
  description: 'MoonshotAI Kimi K2.6 via OpenRouter. Multimodal long-horizon coding and multi-agent orchestration model with 262K context.',
3597
3727
  },
3728
+ {
3729
+ id: 'moonshotai/kimi-k2.7-code',
3730
+ aliases: ['Kimi K2.7 Code', 'kimi-k2.7-code', 'kimi-k2-7-code', 'kimi-k2.7', 'kimi-k2-7'],
3731
+ provider: 'openrouter',
3732
+ openrouter_id: 'moonshotai/kimi-k2.7-code',
3733
+ cost: {
3734
+ input_per_million: 0.74,
3735
+ output_per_million: 3.5,
3736
+ cached_input_per_million: 0.15,
3737
+ },
3738
+ features: {
3739
+ context_length: 262144,
3740
+ input_modality: ['text', 'image'],
3741
+ output_modality: ['text'],
3742
+ tool_use: true,
3743
+ streaming: true,
3744
+ json_output: true,
3745
+ max_output_tokens: 16384,
3746
+ reasoning_output: true,
3747
+ },
3748
+ class: 'code',
3749
+ description: 'MoonshotAI Kimi K2.7 Code via OpenRouter. Coding-focused long-context model for software engineering and agentic coding workflows.',
3750
+ },
3598
3751
  ];
3599
3752
  export function findModel(modelId) {
3600
3753
  const externalModel = getExternalModel(modelId);