Skip to content

Commit 09ca978

Browse files
sxjeruarvinxx
sxjeru
and
arvinxx
authored
💄 style: Support Hunyuan A13B thinking model (#8278)
* Update novita.ts * feat: 添加新的混元模型并更新处理负载逻辑 * fix * free Gemini 2.5 Pro * 添加 ERNIE 4.5 300B A47B 模型到 siliconcloudChatModels * novita * 添加 GLM-4.1V-Thinking 系列模型到 zhipuChatModels，并更新现有模型的上下文窗口和最大输出设置 * 更新 zhipuChatModels，添加视觉能力并启用新模型；修改 ZhiPu 的检查模型 ID * 移除多个 siliconcloud 废弃模型 * fix * 移除 groq Qwen QwQ 32B 模型配置 * 更新 siliconcloud 模型 * 更新 novita/qwen 模型 * update siliconcloud model * add Pangu Pro MoE 72B A16B * Update novita.ts * Update novita.ts * update novita --------- Co-authored-by: Arvin Xu <arvinx@foxmail.com>
1 parent 1696bf1 commit 09ca978

File tree

9 files changed

+349
-232
lines changed

9 files changed

+349
-232
lines changed

‎src/config/aiModels/google.ts

Lines changed: 15 additions & 26 deletions
Original file line numberDiff line numberDiff line change
@@ -11,7 +11,8 @@ const googleChatModels: AIChatModelCard[] = [
1111
contextWindowTokens: 1_048_576 + 65_536,
1212
description:
1313
'Gemini 2.5 Pro 是 Google 最先进的思维模型，能够对代码、数学和STEM领域的复杂问题进行推理，以及使用长上下文分析大型数据集、代码库和文档。',
14-
displayName: 'Gemini 2.5 Pro (Paid)',
14+
displayName: 'Gemini 2.5 Pro',
15+
enabled: true,
1516
id: 'gemini-2.5-pro',
1617
maxOutput: 65_536,
1718
pricing: {
@@ -77,30 +78,6 @@ const googleChatModels: AIChatModelCard[] = [
7778
},
7879
type: 'chat',
7980
},
80-
{
81-
abilities: {
82-
functionCall: true,
83-
reasoning: true,
84-
search: true,
85-
vision: true,
86-
},
87-
contextWindowTokens: 1_048_576 + 65_536,
88-
description:
89-
'Gemini 2.5 Pro Experimental 是 Google 最先进的思维模型，能够对代码、数学和STEM领域的复杂问题进行推理，以及使用长上下文分析大型数据集、代码库和文档。',
90-
displayName: 'Gemini 2.5 Pro Experimental 03-25',
91-
id: 'gemini-2.5-pro-exp-03-25',
92-
maxOutput: 65_536,
93-
pricing: {
94-
input: 0,
95-
output: 0,
96-
},
97-
releasedAt: '2025-03-25',
98-
settings: {
99-
searchImpl: 'params',
100-
searchProvider: 'google',
101-
},
102-
type: 'chat',
103-
},
10481
{
10582
abilities: {
10683
functionCall: true,
@@ -206,7 +183,7 @@ const googleChatModels: AIChatModelCard[] = [
206183
search: true,
207184
vision: true,
208185
},
209-
contextWindowTokens: 65_536 + 65_536,
186+
contextWindowTokens: 1_048_576 + 65_536,
210187
description:
211188
'Gemini 2.5 Flash-Lite Preview 是 Google 最小、性价比最高的模型，专为大规模使用而设计。',
212189
displayName: 'Gemini 2.5 Flash-Lite Preview 06-17',
@@ -493,6 +470,18 @@ const googleChatModels: AIChatModelCard[] = [
493470
},
494471
type: 'chat',
495472
},
473+
{
474+
contextWindowTokens: 2048 + 8192,
475+
displayName: 'Gemma 3n E2B',
476+
id: 'gemma-3n-e2b-it',
477+
maxOutput: 2048,
478+
pricing: {
479+
cachedInput: 0,
480+
input: 0,
481+
output: 0,
482+
},
483+
type: 'chat',
484+
},
496485
{
497486
contextWindowTokens: 2048 + 8192,
498487
displayName: 'Gemma 3n E4B',

‎src/config/aiModels/groq.ts

Lines changed: 0 additions & 16 deletions
Original file line numberDiff line numberDiff line change
@@ -47,22 +47,6 @@ const groqChatModels: AIChatModelCard[] = [
4747
},
4848
type: 'chat',
4949
},
50-
{
51-
abilities: {
52-
functionCall: true,
53-
reasoning: true,
54-
},
55-
contextWindowTokens: 131_072,
56-
displayName: 'Qwen QwQ 32B',
57-
enabled: true,
58-
id: 'qwen-qwq-32b',
59-
maxOutput: 131_072,
60-
pricing: {
61-
input: 0.29,
62-
output: 0.39,
63-
},
64-
type: 'chat',
65-
},
6650
{
6751
abilities: {
6852
reasoning: true,

‎src/config/aiModels/hunyuan.ts

Lines changed: 79 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -2,6 +2,25 @@ import { AIChatModelCard } from '@/types/aiModel';
22

33
// https://cloud.tencent.com/document/product/1729/104753
44
const hunyuanChatModels: AIChatModelCard[] = [
5+
{
6+
abilities: {
7+
reasoning: true,
8+
search: true,
9+
},
10+
contextWindowTokens: 256_000,
11+
description:
12+
'混元第一个混合推理模型，hunyuan-standard-256K 的升级版本，总参数80B，激活13B，默认是慢思考模式，支持通过参数或者指令进行快慢思考模式切换，慢快思考切换方式为 query 前加/ no_think；整体能力相对上一代全面提升，特别数学、科学、长文理解和 Agent 能力提升显著。',
13+
displayName: 'Hunyuan A13B',
14+
enabled: true,
15+
id: 'hunyuan-a13b',
16+
maxOutput: 32_000,
17+
releasedAt: '2025-06-25',
18+
settings: {
19+
extendParams: ['enableReasoning'],
20+
searchImpl: 'params',
21+
},
22+
type: 'chat',
23+
},
524
{
625
abilities: {
726
reasoning: true,
@@ -366,7 +385,6 @@ const hunyuanChatModels: AIChatModelCard[] = [
366385
contextWindowTokens: 8000,
367386
description: '混元最新多模态模型，支持多语种作答，中英文能力均衡。',
368387
displayName: 'Hunyuan Standard Vision',
369-
enabled: true,
370388
id: 'hunyuan-standard-vision',
371389
maxOutput: 2000,
372390
releasedAt: '2024-12-31',
@@ -403,6 +421,25 @@ const hunyuanChatModels: AIChatModelCard[] = [
403421
releasedAt: '2025-05-26',
404422
type: 'chat',
405423
},
424+
{
425+
abilities: {
426+
reasoning: true,
427+
vision: true,
428+
},
429+
contextWindowTokens: 40_000,
430+
description:
431+
'混元最新版t1-vision多模态理解深度思考模型，支持多模态原生长思维链，相比上一代默认版本模型全面提升。',
432+
displayName: 'Hunyuan T1 Vision 20250619',
433+
id: 'hunyuan-t1-vision-20250619',
434+
maxOutput: 24_000,
435+
pricing: {
436+
currency: 'CNY',
437+
input: 1,
438+
output: 4,
439+
},
440+
releasedAt: '2025-06-19',
441+
type: 'chat',
442+
},
406443
{
407444
abilities: {
408445
reasoning: true,
@@ -414,9 +451,50 @@ const hunyuanChatModels: AIChatModelCard[] = [
414451
displayName: 'Hunyuan T1 Vision',
415452
id: 'hunyuan-t1-vision',
416453
maxOutput: 24_000,
454+
pricing: {
455+
currency: 'CNY',
456+
input: 1,
457+
output: 4,
458+
},
417459
releasedAt: '2025-05-16',
418460
type: 'chat',
419461
},
462+
{
463+
abilities: {
464+
vision: true,
465+
},
466+
contextWindowTokens: 32_000,
467+
description:
468+
'混元最新版turbos-vision视觉语言旗舰大模型，在图文理解相关的任务上，包括基于图片的实体识别、知识问答、文案创作、拍照解题等上面相比上一代默认版本模型全面提升。',
469+
displayName: 'Hunyuan TurboS Vision 20250619',
470+
id: 'hunyuan-turbos-vision-20250619',
471+
maxOutput: 16_000,
472+
pricing: {
473+
currency: 'CNY',
474+
input: 3,
475+
output: 9,
476+
},
477+
releasedAt: '2025-06-19',
478+
type: 'chat',
479+
},
480+
{
481+
abilities: {
482+
vision: true,
483+
},
484+
contextWindowTokens: 32_000,
485+
description:
486+
'此模型适用于图文理解场景，是基于混元最新 turbos 的新一代视觉语言旗舰大模型，聚焦图文理解相关任务，包括基于图片的实体识别、知识问答、文案创作、拍照解题等方面，相比前一代模型全面提升。',
487+
displayName: 'Hunyuan TurboS Vision',
488+
id: 'hunyuan-turbos-vision',
489+
maxOutput: 24_000,
490+
pricing: {
491+
currency: 'CNY',
492+
input: 3,
493+
output: 9,
494+
},
495+
releasedAt: '2025-05-23',
496+
type: 'chat',
497+
},
420498
{
421499
abilities: {
422500
vision: true,

‎src/config/aiModels/novita.ts

Lines changed: 50 additions & 56 deletions
Original file line numberDiff line numberDiff line change
@@ -2,6 +2,33 @@ import { AIChatModelCard } from '@/types/aiModel';
22

33
// https://novita.ai/pricing
44
const novitaChatModels: AIChatModelCard[] = [
5+
{
6+
abilities: {
7+
reasoning: true,
8+
},
9+
contextWindowTokens: 131_072,
10+
displayName: 'Kimi K2 Instruct',
11+
id: 'moonshotai/kimi-k2-instruct',
12+
pricing: {
13+
input: 0.57,
14+
output: 2.3,
15+
},
16+
type: 'chat',
17+
},
18+
{
19+
abilities: {
20+
reasoning: true,
21+
vision: true,
22+
},
23+
contextWindowTokens: 65_536,
24+
displayName: 'GLM 4.1V 9B Thinking',
25+
id: 'thudm/glm-4.1v-9b-thinking',
26+
pricing: {
27+
input: 0.035,
28+
output: 0.138,
29+
},
30+
type: 'chat',
31+
},
532
{
633
contextWindowTokens: 120_000,
734
displayName: 'ERNIE 4.5 0.3B',
@@ -23,12 +50,15 @@ const novitaChatModels: AIChatModelCard[] = [
2350
type: 'chat',
2451
},
2552
{
53+
abilities: {
54+
reasoning: true,
55+
},
2656
contextWindowTokens: 123_000,
2757
displayName: 'ERNIE 4.5 300B A47B Paddle',
2858
id: 'baidu/ernie-4.5-300b-a47b-paddle',
2959
pricing: {
30-
input: 0.3,
31-
output: 1,
60+
input: 0.28,
61+
output: 1.1,
3262
},
3363
type: 'chat',
3464
},
@@ -48,6 +78,7 @@ const novitaChatModels: AIChatModelCard[] = [
4878
},
4979
{
5080
abilities: {
81+
functionCall: true,
5182
reasoning: true,
5283
vision: true,
5384
},
@@ -62,9 +93,10 @@ const novitaChatModels: AIChatModelCard[] = [
6293
},
6394
{
6495
abilities: {
96+
functionCall: true,
6597
reasoning: true,
6698
},
67-
contextWindowTokens: 128_000,
99+
contextWindowTokens: 1_000_000,
68100
displayName: 'MiniMax M1 80K',
69101
id: 'minimaxai/minimax-m1-80k',
70102
pricing: {
@@ -220,6 +252,17 @@ const novitaChatModels: AIChatModelCard[] = [
220252
},
221253
type: 'chat',
222254
},
255+
{
256+
contextWindowTokens: 32_768,
257+
description: 'Gemma 3 1B 是谷歌的一款开源语言模型，以其在效率和性能方面设立了新的标准。',
258+
displayName: 'Gemma 3 1B',
259+
id: 'google/gemma-3-1b-it',
260+
pricing: {
261+
input: 0,
262+
output: 0,
263+
},
264+
type: 'chat',
265+
},
223266
{
224267
contextWindowTokens: 60_288,
225268
description: 'Mistral Nemo 是多语言支持和高性能编程的7.3B参数模型。',
@@ -318,8 +361,8 @@ const novitaChatModels: AIChatModelCard[] = [
318361
enabled: true,
319362
id: 'deepseek/deepseek-v3-0324',
320363
pricing: {
321-
input: 0.33,
322-
output: 1.3,
364+
input: 0.28,
365+
output: 1.14,
323366
},
324367
type: 'chat',
325368
},
@@ -527,55 +570,19 @@ const novitaChatModels: AIChatModelCard[] = [
527570
},
528571
type: 'chat',
529572
},
530-
{
531-
contextWindowTokens: 32_000,
532-
displayName: 'Qwen2.5 7B Instruct',
533-
id: 'qwen/qwen2.5-7b-instruct',
534-
pricing: {
535-
input: 0,
536-
output: 0,
537-
},
538-
type: 'chat',
539-
},
540-
{
541-
abilities: {
542-
functionCall: true,
543-
},
544-
contextWindowTokens: 32_000,
545-
displayName: 'GLM 4 9B 0414',
546-
id: 'thudm/glm-4-9b-0414',
547-
pricing: {
548-
input: 0,
549-
output: 0,
550-
},
551-
type: 'chat',
552-
},
553573
{
554574
abilities: {
555575
functionCall: true,
556576
},
557577
contextWindowTokens: 32_000,
558-
displayName: 'GLM Z1 9B 0414',
559-
id: 'thudm/glm-z1-9b-0414',
578+
displayName: 'Qwen2.5 7B Instruct',
579+
id: 'qwen/qwen2.5-7b-instruct',
560580
pricing: {
561581
input: 0,
562582
output: 0,
563583
},
564584
type: 'chat',
565585
},
566-
{
567-
abilities: {
568-
functionCall: true,
569-
},
570-
contextWindowTokens: 32_000,
571-
displayName: 'GLM Z1 32B 0414',
572-
id: 'thudm/glm-z1-32b-0414',
573-
pricing: {
574-
input: 0.24,
575-
output: 0.24,
576-
},
577-
type: 'chat',
578-
},
579586
{
580587
abilities: {
581588
functionCall: true,
@@ -589,19 +596,6 @@ const novitaChatModels: AIChatModelCard[] = [
589596
},
590597
type: 'chat',
591598
},
592-
{
593-
abilities: {
594-
functionCall: true,
595-
},
596-
contextWindowTokens: 32_000,
597-
displayName: 'GLM Z1 Rumination 32B 0414',
598-
id: 'thudm/glm-z1-rumination-32b-0414',
599-
pricing: {
600-
input: 0.24,
601-
output: 0.24,
602-
},
603-
type: 'chat',
604-
},
605599
];
606600

607601
export const allModels = [...novitaChatModels];

0 commit comments

Comments
 (0)