HyperAI
Command Palette
Search for a command to run...
CritPt v2
CritPt v2 是 CritPt 的修订版，该基准用于测试 LLM 在研究级物理推理任务上的表现，包含 70 个复合研究挑战。模型在离线 Linux 沙箱中借助代码执行工具求解问题。
- 智能指数权重
- 10%
- 模型
- 110 当前，128 总共
- 中位数
- 33.5%
- 范围
- 0.0%–76.5%
| 排名 | 比较 | 模型 | 评分 | 成本 / 任务 | 时间 / 任务 | 输出 token / 任务 |
|---|---|---|---|---|---|---|
| 51 |  MiMo-V2.6-Pro Xiaomi | 34.0% | $0.17 | 34m 21s | 116K | |
| 52 | 33.5% | $0.093 | 48.5s | 6.76K | ||
| 53 |  Gemini 3.8 Flash (Low) Google | 33.5% | — | — | — | |
| 54 |  GLM-5.3 (Max) Z AI | 33.5% | $1.54 | 38m 42s | 159K | |
| 55 | 33.5% | $0.017 | 30.0s | 2.07K | ||
| 56 |  Grok 4.7 (Xhigh) SpaceXAI | 33.5% | $2.24 | 28m 0s | 107K | |
| 57 |  MiniMax-M3.1-Flash MiniMax | 32.5% | $0.17 | 17m 8s | 164K | |
| 58 |  Step 5 Preview StepFun | 32.5% | $0.36 | 8m 6s | 70K | |
| 59 |  GPT-6 Luna (High) OpenAI | 31.0% | $0.0096 | 2m 25s | 14.1K | |
| 60 |  Claude Haiku 5.5 (Medium) Anthropic | 30.5% | $0.017 | 1m 49s | 24.5K | |
| 61 | MiMo-V2.6-Flash Xiaomi | 29.5% | $0.062 | 61m 41s | 130K | |
| 62 | 29.5% | $0.51 | 6m 27s | 66.8K | ||
| 63 |  Qwen3.8 Max (0902) Alibaba | 29.5% | $1.76 | 42m 19s | 130K | |
| 64 | MiniMax-M3 MiniMax | 28.0% | $0.33 | 16m 46s | 125K | |
| 65 |  DeepSeek V4.1 Flash 1017 (High) DeepSeek | 27.5% | $0.18 | 12m 53s | 109K | |
| 66 | GPT-6 Luna (Medium) OpenAI | 27.5% | $0.0055 | 1m 22s | 8.14K | |
| 67 |  Mistral Large 4 (High) Mistral | 27.0% | $0.32 | 11m 42s | 82.3K | |
| 68 |  Beam 1.0 Reflection | 26.5% | $0.089 | 2m 31s | 43.6K | |
| 69 |  Agnes 2.5 Pro Theta Sapiens AI | 26.0% | $0.14 | 31m 48s | 369K | |
| 70 | 26.0% | $0.073 | 36.1s | 5.02K | ||
| 71 | DeepSeek V4.1 Flash 1017 (Max) DeepSeek | 25.0% | $0.20 | 14m 23s | 119K | |
| 72 | GLM-5.3-Flash Z AI | 24.5% | $0.36 | 130m 1s | 386K | |
| 73 | Qwen3.8-Flash-Next Alibaba | 24.5% | — | — | — | |
| 74 | DeepSeek V4 Pro 0813 (Max) DeepSeek | 23.5% | $0.89 | 34m 28s | 143K | |
| 75 | Mistral Large 4 (Medium) Mistral | 22.5% | $0.20 | 7m 37s | 51.6K | |
| 76 | Claude Haiku 5.5 (Low) Anthropic | 21.5% | $0.0083 | 56.1s | 12.3K | |
| 77 | GLM-5.3 (Low) Z AI | 21.5% | $0.76 | 21m 37s | 76.6K | |
| 78 |  Inkling Small Thinking Machines | 18.5% | $0.14 | 15m 30s | 66.6K | |
| 79 | Inkling (Xhigh) Thinking Machines | 18.0% | $0.55 | 13m 59s | 81.7K | |
| 80 |  Solar Pro 4 Upstage | 18.0% | — | — | — | |
| 81 |  Apodex 1.1 Apodex | 17.0% | $0.49 | 15m 44s | 134K | |
| 82 | Qwen3.8 27B (Medium) Alibaba | 16.5% | $0.75 | 43m 33s | 117K | |
| 83 |  Nanbeige4.3-33B Nanbeige | 15.5% | — | — | 161K | |
| 84 | 15.0% | $0.36 | 6m 12s | 70.2K | ||
| 85 |  Nemotron 3.5 Super VL NVIDIA | 13.5% | $0.12 | 6m 44s | 105K | |
| 86 |  K2 Horizon 375B A23B Institute of Foundation Models | 12.0% | $0 | 9m 0s | 102K | |
| 87 | GPT-6 Luna (Low) OpenAI | 9.5% | $0.0019 | 31.6s | 2.74K | |
| 88 | 8.5% | $0.42 | 17m 9s | 68.3K | ||
| 89 | K2 Horizon MoVA 36B A4B Institute of Foundation Models | 8.0% | — | — | — | |
| 90 | 7.0% | $0.07 | 6m 44s | 36.2K | ||
| 91 | Qwen3.8 27B (Low) Alibaba | 5.5% | $0.42 | 23m 49s | 71.7K | |
| 92 | 5.0% | $0.0011 | 15.7s | 1.42K | ||
| 93 | Gemma 4 31B (Reasoning) Google | 4.0% | — | — | — | |
| 94 | gpt-oss-120b (High) OpenAI | 4.0% | $0.054 | 7m 35s | 73.5K | |
| 95 |  K-EXAONE 2.0 LG AI Research | 4.0% | — | — | 75K | |
| 96 | Qwen3.8 27B (Xhigh) Alibaba | 4.0% | $0.52 | 27m 32s | 76.2K | |
| 97 | Gemini 3.5 Flash-Lite Google | 3.5% | $0.18 | 1m 40s | 30.9K | |
| 98 |  Command A+ Cohere | 2.5% | $0 | 5m 13s | 62.4K | |
| 99 | Qwen3.8 27B (Non-reasoning) Alibaba | 2.0% | $0.29 | 12m 47s | 31.2K | |
| 100 | 1.5% | — | — | — |
