Mathematical Reasoning on AIME 24 (accuracy, Avg.)
93.33AIME 24 AccuracyMix-Quant
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Mix-QuantBase Model=Gemma-4-31B-it2026.05 | 93.33 | 81.39 | |
| Gemma-4-31B-itQuantization=BF162026.05 | 92.22 | 82.36 | |
| Gemma-4-31B-it-NVFP4Quantization=NVFP42026.05 | 83.33 | 78.26 | |
| Mix-QuantBase Model=Gemma-4-26B-A4B-it2026.05 | 78.67 | 71.93 | |
| Gemma-4-26B-A4B-itQuantization=BF162026.05 | 77.67 | 71.94 | |
| Mix-QuantBase Model=Qwen3-8B2026.05 | 76.66 | 61 | |
| Qwen3-8BQuantization=BF162026.05 | 75.54 | 62.11 | |
| Gemma-4-26B-A4B-it-NVFP4Quantization=NVFP42026.05 | 75.33 | 66.31 | |
| Mix-QuantBase Model=Qwen3.5-9B2026.05 | 70.33 | 70.59 | |
| Qwen3.5-9BQuantization=BF162026.05 | 68.89 | 72.04 | |
| Qwen3-8B-NVFP4Quantization=NVFP42026.05 | 66.53 | 55.22 | |
| Qwen3.5-9B-NVFP4Quantization=NVFP42026.05 | 54.44 | 63.26 |