Math on LiC Math
92.9Concat ScoreLlama4-Scout
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Llama4-Scout2025.10 | 92.9 | 67 | 72.1 | — | |
| DeepSeek-R12025.10 | 92.9 | 67.3 | 72.4 | — | |
| GPT-4o2025.10 | 91.9 | 67.9 | 73.9 | 39.9 | |
| Llama3.3-70B2025.10 | 91.8 | 61.5 | 67 | — | |
| Phi-42025.10 | 90.4 | 52.5 | 58.1 | — | |
| GPT-4.12025.10 | 89.7 | 70.7 | 78.8 | 40.5 | |
| Gemini-2.5-Pro2025.10 | 89.3 | 64.3 | 72 | 46.2 | |
| Gemini-2.5-Flash2025.10 | 88.4 | 66.1 | 74.8 | 47 | |
| Qwen3-8Bsource=Grattafiori et al.2025.10 | 88.4 | 72.8 | 82.4 | 39.1 | |
| GPT4o-mini2025.10 | 88 | 58.7 | 66.7 | — | |
| Qwen3-8BFinetuning=SFT, Distillation=Gemini-2.5-Pro2025.10 | 87.4 | 75.6 | 86.5 | 57.4 | |
| Claude3.7-Sonnet2025.10 | 87.2 | 70 | 80.3 | — | |
| Qwen3-8BOptimization=RLAAR2025.10 | 87.2 | 82.2 | 94.3 | 80.8 | |
| Qwen2.5-7BOptimization=RLAAR2025.10 | 83.4 | 79.2 | 95 | 77 | |
| Qwen2.5-7BFinetuning=SFT, Distillation=Gemini-2.5-Pro2025.10 | 82.7 | 60.3 | 72.9 | 57.2 | |
| Qwen2.5-7B2025.10 | 82.2 | 59.8 | 72.8 | 36.1 | |
| OLMo22025.10 | 80.1 | 46.3 | 57.8 | — | |
| GPT-o32025.10 | 80 | 63.1 | 78.9 | — | |
| Claude3-Haiku2025.10 | 76.1 | 47.1 | 61.9 | — | |
| Qwen3-1.7BOptimization=RLAAR2025.10 | 70.9 | 60 | 84.7 | 77.2 | |
| Qwen3-1.7Bsource=Grattafiori et al.2025.10 | 67.6 | 44.4 | 65.7 | 26.8 |