Math on GSME (test)
68.76Surrogate Accuracy (Initial)GPT 5
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| GPT 5Target Model=GPT 52026.01 | 68.76 | 76.2 | — | — | 361 | 0.35 | — | — | |
| Gemini 2.5 FlashTarget Model=Gemini 2.5 Flash2026.01 | 68.23 | 78.45 | — | — | 167 | 0.35 | — | — | |
| Gemini 2.5 ProTarget Model=Gemini 2.5 Pro2026.01 | 62.29 | 66.92 | — | — | 265 | 0.35 | — | — | |
| o4-miniTarget Model=o4-mini2026.01 | 55.4 | 69.45 | — | — | 768 | 0.35 | — | — |