Tool-integrated Mathematical Reasoning on AIME 24
76Pass@1Qwen3-14B + PRUNETIR
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Qwen3-14B + PRUNETIRPRUNETIR=true2026.05 | 76 | 3.7 | 8.8 | |
| Qwen3-8B + PRUNETIRPRUNETIR=true2026.05 | 72.7 | 4.2 | 9.5 | |
| Qwen3-14BPRUNETIR=false2026.05 | 71.7 | 4.4 | 9.7 | |
| ReTool-32B + PRUNETIRPRUNETIR=true2026.05 | 69.7 | 6.6 | 5 | |
| ReTool-32BOfficial Release=true, PRUNETIR=false2026.05 | 67 | — | — | |
| START-32BOfficial Release=true2026.05 | 66.7 | — | — | |
| Qwen3-8BPRUNETIR=false2026.05 | 62.1 | 7.7 | 11.5 | |
| SimpleTIR-32BOfficial Release=true2026.05 | 59.9 | — | — | |
| OpenAI o1-previewOfficial Release=true2026.05 | 44.6 | — | — |