List your company

Math Proof Reward Modeling on OPC (test)

83.1Accuracyo3
68.85272.55176.2579.949Feb 2, 2026

Evaluation results

MethodLinks
o3Access=Closed-source2026.0283.1
Gemini-2.5-FlashAccess=Closed-source2026.0282.7
ProofRMSize=32B2026.0281.3
Deepseek-R12026.0280.9
ProofRMSize=14B2026.0277.4
ProofRMSize=8B2026.0275.7
Distill-Qwen3Size=8B2026.0270.8
Distill-Qwen3Size=32B2026.0270.2
Distill-Qwen3Size=14B2026.0269.4