context.vn
Menu

pinch Bench

44 models evaluated

#ModelProviderTypeScore
1Anthropic93.5%anthropic/claude-opus-4.8-fast
2Qwen3.7 MaxAlibabaProprietary92.5%
3Claude Opus 4.8AnthropicProprietary90.5%
5MiMo-V2.5Xiaomi · Closed89.7%
6Grok Build 0.1xAI · Closed88.9%
7GPT-5.6 LunaOpenAI · Closed88.7%
9MiMo-V2.5-ProXiaomi · Closed87.5%
10GLM-5.2Z.AI · Open weight87.0%
12GPT-5.6 SolOpenAI · Closed84.2%
14DeepSeek V4 Flash 0731DeepSeek · Closed81.7%
15Gemini 3.1 ProGoogle · Closed81.0%
16Gemini 3.1 Flash-LiteGoogle · Closed80.5%
17Grok 4.20xAI · Closed80.3%
18Step 3.5 FlashStepFun · Open weight79.4%
19GPT-5.4 miniOpenAI · Closed79.2%
20Kimi K2.7 CodeMoonshot AI · Open weight76.1%
21Claude Opus 4.7Anthropic · Closed76.0%
22GPT-5.6 TerraOpenAI · Closed75.9%
23GPT-5.4OpenAI · Closed75.7%
24GPT-5.5OpenAI · Closed75.5%
25Grok 4.5xAI · Closed75.2%
26Seed-2.0-LiteByteDance · Closed75.0%
27Gemini 3.5 FlashGoogle · Closed74.2%
28Grok 4.3xAI · Closed73.7%
29Qwen3.6 PlusAlibaba · Closed72.5%
30Gemini 3 FlashGoogle · Closed72.1%
31GLM-5-TurboZ.AI · Closed71.8%
32Claude Opus 4.6Anthropic · Closed69.9%
35GPT-5.4 nanoOpenAI · Closed69.0%
36Claude Haiku 4.5Anthropic · Closed67.7%
37GLM-5V-TurboZ.AI · Closed67.6%
38MiniMax M2.7MiniMax · Open weight66.8%
39Trinity-Large-PreviewArcee AI · Open weight65.7%
41Claude Sonnet 4.6Anthropic · Closed62.7%
43DeepSeek V4 Pro 0813DeepSeek · Closed61.1%
44GLM-5.1Z.AI · Open weight59.9%
46Claude Fable 5Anthropic · Closed54.8%
47Kimi K2.5Moonshot AI · Open weight54.6%
52GPT-OSS 120BOpenAI · Open weight44.8%
53Nemotron 3 Super 120B A12BNVIDIA · Open weight42.2%
54Mercury 2Inception · Closed39.6%
56GPT-OSS 20BOpenAI · Open weight36.3%
57GPT-5.5 ProOpenAI · Closed21.4%
59Llama 4 ScoutMeta · Open weight3.2%