Mathematical Reasoning on RealMath (200 held-out questions)
85.1AccuracyMAGE
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MAGEBackbone=Qwen3-8B, Judge=Claude Sonnet 4.6, MAGE Executor Model=Sonnet2026.05 | 85.1 | 31.6 | |
| MAGEBackbone=Qwen3-8B, Judge=Claude Sonnet 4.6, MAGE Executor Model=Opus2026.05 | 82.8 | 31.6 | |
| 0-shot CoTBackbone=Qwen3-8B, Judge=Claude Sonnet 4.6, Prompting Scaffold=0-shot CoT2026.05 | 53.5 | — | |
| ReActBackbone=Qwen3-8B, Judge=Claude Sonnet 4.6, Prompting Scaffold=ReAct2026.05 | 43.5 | — | |
| ReflexionBackbone=Qwen3-8B, Judge=Claude Sonnet 4.6, Prompting Scaffold=Reflexion2026.05 | 34.5 | — | |
| SC10Backbone=Qwen3-8B, Judge=Claude Sonnet 4.6, Prompting Scaffold=SC10, Samples=102026.05 | 33.5 | — | |
| 8-shotBackbone=Qwen3-8B, Judge=Claude Sonnet 4.6, Prompting Scaffold=8-shot2026.05 | 32.5 | — |