Science Reasoning on GPQA (Score)
81.4ScoreDeepSeek v3.2
Evaluation Results
| Method | Links | |
|---|---|---|
| DeepSeek v3.22025.12 | 81.4 | |
| GLM-4.62025.12 | 78.8 | |
| DeepSeek R1 05282025.12 | 77.5 | |
| GPT-OSSParameters=120B2025.12 | 77.3 | |
| GLM-4.52025.12 | 77 | |
| INTELLECT-3Parameters=100B+2025.12 | 74.4 | |
| GLM-4.5-Air2025.12 | 73.3 | |
| TPOModel=LLaMA-3.2-3B2025.05 | 39 | |
| GRPOModel=LLaMA-3.2-3B2025.05 | 38 | |
| CPOModel=LLaMA-3.2-3B2025.05 | 35.5 | |
| KTOModel=LLaMA-3.2-3B2025.05 | 35 | |
| TI-DPOModel=LLaMA-3.2-3B2025.05 | 34.5 | |
| DPOModel=LLaMA-3.2-3B2025.05 | 34 | |
| TDPOModel=LLaMA-3.2-3B2025.05 | 34 | |
| SIMPOModel=LLaMA-3.2-3B2025.05 | 33.5 | |
| SFTModel=LLaMA-3.2-3B2025.05 | 33 | |
| IPOModel=LLaMA-3.2-3B2025.05 | 31 |