Mathematical Reasoning on AIME 2025-2026 (test)
93.3AccuracyT3-Reflect
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| T3-ReflectModel=GPT-52026.05 | 93.3 | 7.6 | |
| GitHubModel=GPT-52026.05 | 91.7 | 5.8 | |
| Chunked traj.Model=GPT-52026.05 | 91.7 | 5.8 | |
| T3-StructModel=GPT-52026.05 | 91.7 | 5.8 | |
| Wikipedia-RPJModel=GPT-52026.05 | 90 | 3.8 | |
| Wikipedia-DPRModel=GPT-52026.05 | 88.3 | 1.8 | |
| CompactDSModel=GPT-52026.05 | 88.3 | 1.8 | |
| T3-SemanticModel=GPT-52026.05 | 88.3 | 1.8 | |
| No RAGModel=GPT-52026.05 | 86.7 | — | |
| Full traj.Model=GPT-52026.05 | 86.7 | 0 | |
| OpenWebMathModel=GPT-52026.05 | 85 | -2 | |
| ArxivModel=GPT-52026.05 | 85 | -2 | |
| Chunked traj.Model=GPT-OSS 120B2026.05 | 85 | 8.6 | |
| StackExchangeModel=GPT-52026.05 | 83.3 | -3.9 | |
| Tavily Search APIModel=GPT-52026.05 | 83.3 | -3.9 | |
| T3-SemanticModel=GPT-OSS 120B2026.05 | 83.3 | 6.4 | |
| T3-SemanticModel=Gemini-2.5 Flash2026.05 | 83.3 | 56.3 | |
| T3-StructModel=GPT-OSS 120B2026.05 | 81.7 | 4.3 | |
| T3-ReflectModel=GPT-OSS 120B2026.05 | 81.7 | 4.3 | |
| CompactDSModel=GPT-OSS 120B2026.05 | 80 | 2.2 | |
| Chunked traj.Model=Gemini-2.5 Flash2026.05 | 80 | 50.1 | |
| No RAGModel=GPT-OSS 120B2026.05 | 78.3 | — | |
| ArxivModel=GPT-OSS 120B2026.05 | 78.3 | 0 | |
| StackExchangeModel=GPT-OSS 120B2026.05 | 76.7 | -2 | |
| Wikipedia-RPJModel=GPT-OSS 120B2026.05 | 76.7 | -2 | |
| GitHubModel=GPT-OSS 120B2026.05 | 76.7 | -2 | |
| T3-ReflectModel=Gemini-2.5 Flash2026.05 | 76.7 | 43.9 | |
| Tavily Search APIModel=GPT-OSS 120B2026.05 | 75 | -4.2 | |
| Full traj.Model=GPT-OSS 120B2026.05 | 73.3 | -6.4 | |
| Full traj.Model=Gemini-2.5 Flash2026.05 | 73.3 | 37.5 | |
| T3-StructModel=Gemini-2.5 Flash2026.05 | 73.3 | 37.5 | |
| Wikipedia-DPRModel=GPT-OSS 120B2026.05 | 71.7 | -8.4 | |
| OpenWebMathModel=GPT-OSS 120B2026.05 | 63.3 | -19.2 | |
| GitHubModel=Gemini-2.5 Flash2026.05 | 60 | 12.6 | |
| Tavily Search APIModel=Gemini-2.5 Flash2026.05 | 60 | 12.6 | |
| CompactDSModel=Gemini-2.5 Flash2026.05 | 58.3 | 9.4 | |
| Wikipedia-DPRModel=Gemini-2.5 Flash2026.05 | 56.7 | 6.4 | |
| No RAGModel=Gemini-2.5 Flash2026.05 | 53.3 | — | |
| ArxivModel=Gemini-2.5 Flash2026.05 | 51.7 | -3 | |
| StackExchangeModel=Gemini-2.5 Flash2026.05 | 46.7 | -12.4 | |
| OpenWebMathModel=Gemini-2.5 Flash2026.05 | 45 | -15.6 | |
| Wikipedia-RPJModel=Gemini-2.5 Flash2026.05 | 41.7 | -21.8 |