Math on Math OOD
92.04AccuracyGPT-5.2-chat-latest
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GPT-5.2-chat-latest2026.05 | 92.04 | 88.5 | |
| GPT-5-chat-latest2026.05 | 90.27 | 71.68 | |
| Qwen-3-14B2026.05 | 88.5 | 62.83 | |
| GRPO+AIR2026.05 | 88.49 | 61.06 | |
| GSPO2026.05 | 87.61 | 54.87 | |
| GSPO+AIR2026.05 | 86.28 | 51.33 | |
| Llama-4-Maverick2026.05 | 84.51 | 46.02 | |
| GSPO+V-REx2026.05 | 84.07 | 45.13 | |
| Gemini-3-Pro-Preview2026.05 | 83.63 | 43.36 | |
| GRPO+V-REx2026.05 | 83.19 | 42.48 | |
| GRPO2026.05 | 82.3 | 40.71 | |
| Qwen-2.5-14B-Instruct2026.05 | 81.86 | 25.66 | |
| Gemini-2.5-Pro2026.05 | 79.2 | 20.35 |