Mathematics Reasoning on AIME 2025 (Accuracy and Mean Output Tokens)
78.3AccuracyApriel-Reasoner (Ours)
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Apriel-Reasoner (Ours)Size=15B2026.04 | 78.3 | 11,300 | |
| Nemotron-CascadeSize=14B2026.04 | 76 | 19,000 | |
| Apriel-BaseSize=15B2026.04 | 73.3 | 16,600 | |
| Apriel-Base + RLVR w/ LPSize=15B, Length Penalty (LP)=true2026.04 | 71.7 | 11,100 | |
| Qwen3Size=14B2026.04 | 68 | 16,900 | |
| Phi-4-reasoningSize=14B2026.04 | 57.3 | 12,500 |