AI Flash Report

HLE leaderboard

52 models ranked, highest score first.

HLE leaderboard — 52 models ranked by score
# Model Company Score
1 Claude Fable 5 Anthropic 53.3%
2 GLM-5 Zhipu AI 50.4%
3 Claude Opus 4.8 Anthropic 45.7%
4 GPT-5.5 OpenAI 44.3%
5 GPT-5.3 Codex OpenAI 39.9%
6 Qwen3.7 Max Alibaba 38.1%
7 MiniMax-M3 MiniMax 37.1%
8 DeepSeek V4 Pro DeepSeek 35.9%
9 GPT-5.2 OpenAI 35.4%
10 Grok 4.3 xAI 35.0%
11 MiMo-V2.5-Pro Xiaomi 33.8%
12 Qwen3.7 Plus Alibaba 33.4%
13 DeepSeek V4 Flash DeepSeek 32.1%
14 Qwen3.6 Max Preview Alibaba 28.9%
15 Claude Opus 4.5 Anthropic 28.4%
16 GPT-5.1 OpenAI 26.5%
17 Hy3-preview Tencent 25.5%
18 MiMo-V2.5 Xiaomi 25.2%
19 Gemini 3.5 Flash Google 23.1%
20 DeepSeek V3.2 DeepSeek 22.2%
21 Qwen3.6 27B Alibaba 21.6%
22 GPT-5.5 Instant OpenAI 20.3%
23 Step 3.7 Flash StepFun 19.9%
24 Ring-2.6-1T InclusionAI 18.3%
25 HyperNova 60B 2605 Multiverse Computing 15.1%
26 Gemma 4 12B Google 14.8%
27 Mistral Medium 3.5 Mistral 12.8%
28 Gemini 2.5 Flash Google 11.1%
29 Claude Sonnet 4.6 Anthropic 10.8%
30 North Mini Code Cohere 9.9%
31 Ling-2.6-1T InclusionAI 8.2%
32 LFM2.5-8B-A1B Liquid AI 6.9%
33 MiniCPM5-1B OpenBMB 6.5%
34 Ling 2.6 Flash InclusionAI 6.2%
35 JT-35B-Flash China Mobile 6.1%
36 GPT-5 OpenAI 5.4%
37 MiniCPM-V 4.6 1.3B OpenBMB 4.9%
38 Gemini 2.0 Flash Google 4.7%
39 Granite 4.1 30B IBM 4.2%
40 Claude 2.1 Anthropic 4.2%
41 Mistral Large 3 Mistral 4.1%
42 Claude 3 Haiku Anthropic 3.9%
43 Gemini 1.5 Pro Google 3.9%
44 Granite 4.1 8B IBM 3.8%
45 Grok-2 xAI 3.8%
46 Claude 3 Sonnet Anthropic 3.8%
47 Claude 3.5 Sonnet Anthropic 3.7%
48 DeepSeek-V3 DeepSeek 3.6%
49 Granite 4.1 3B IBM 3.4%
50 Mistral Large Mistral 3.4%
51 GPT-4 Turbo OpenAI 3.3%
52 Claude 3 Opus Anthropic 3.1%