HLE
OpenCompass · 48 models
| Rank | Model | Score |
|---|---|---|
| 1 | Gemini-3-Pro-Preview | 38 |
| 2 | DeepSeek-V3.2-Speciale | 28.6 |
| 2 | Kimi-K2.5 | 28.6 |
| 4 | GPT-5-2025-08-07 (high) | 28.5 |
| 5 | GLM-5 | 28.1 |
| 6 | Qwen3.5-397B-A17B | 27.5 |
| 7 | GLM-4.7 | 25.4 |
| 8 | DeepSeek-V3.2 | 23.2 |
| 9 | MiniMax-M2.5 | 22.2 |
| 10 | Step-3.5-Flash | 21.6 |
| 11 | GPT-5-mini-2025-08-07 (high) | 21.5 |
| 12 | Kimi-K2-Thinking | 21.3 |
| 12 | o3-2025-04-16 | 21.3 |
| 14 | Gemini-2.5-Pro | 21.1 |
| 15 | MiMo-V2-Flash | 20.5 |
| 16 | TeleChat-thinking | 20.3 |
| 17 | GLM-4.6 | 19.3 |
| 18 | Qwen3-235B-A22B-Thinking-2507 | 18.5 |
| 19 | gpt-oss-120b (high) | 18.3 |
| 20 | GLM-4.5 | 16.9 |
| 21 | DeepSeek-R1-0528 | 14.4 |
| 21 | JT-Think | 14.4 |
| 23 | o4-mini-2025-04-16 (high) | 14.3 |
| 24 | Qwen3-Next-80B-A3B-Thinking | 13.5 |
| 24 | doubao-seed-1.6-thinking-250615 | 13.5 |
| 26 | MiniMax-M2 | 13.4 |
| 27 | Qwen3-235B-A22B-Instruct-2507 | 12.3 |
| 28 | Qwen3-30B-A3B-Thinking-2507 | 11.7 |
| 29 | gpt-oss-20b (high) | 11.6 |
| 30 | GPT-5-nano-2025-08-07 (high) | 11 |
| 31 | Seed-OSS-36B-Instruct | 9.5 |
| 32 | MiniCPM-SALA | 9.3 |
| 33 | Claude Sonnet 4 (Thinking) | 8.7 |
| 34 | LongCat-Flash-Chat | 8.5 |
| 34 | Qwen3-32B (Thinking) | 8.5 |
| 36 | Qwen3-Next-80B-A3B-Instruct | 8 |
| 37 | Qwen3-30B-A3B-Instruct-2507 | 7.7 |
| 38 | Finix-P1-32B (Thinking) | 6.9 |
| 39 | Kimi-K2-Instruct | 6.7 |
| 40 | ERNIE-4.5-21B-A3B-Thinking | 6.5 |
| 41 | GPT-4.1-20250414 | 6 |
| 41 | Hunyuan-A13B-Instruct | 6 |
| 41 | Qwen3-4B-Thinking-2507 | 6 |
| 44 | Qwen3-8B (Thinking) | 5.5 |
| 45 | Llama4-Maverick-17B-128E-Instruct | 5.3 |
| 46 | Qwen3-4B-Instruct-2507 | 5.1 |
| 47 | Gemma-3-27B-it | 4.2 |
| 48 | ERNIE-4.5-300B-A47B-PT | 3.8 |