Multimodal Autoformalization on MATHVERSE Solid Geometry
80Compilation SuccessGemini-3-Pro
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Gemini-3-ProModality=Image2026.01 | 80 | 80 | 50 | |
| Gemini-3-ProModality=Text2026.01 | 70 | 70 | — | |
| Gemini-3-ProModality=Image2026.01 | 52 | 52 | 52 | |
| Gemini-2.5-ProModality=Image2026.01 | 32 | 24 | 24 | |
| Gemini-2.5-ProModality=Image2026.01 | 30 | 10 | 10 | |
| Gemini-2.5-ProModality=Text2026.01 | 30 | 20 | — | |
| GPT-5Modality=Image2026.01 | 28 | 28 | 28 | |
| Qwen3-VL-235BModality=Image2026.01 | 20 | 16 | 12 | |
| GPT-5Modality=Text2026.01 | 20 | 10 | — | |
| Qwen2.5-VL-72BModality=Image2026.01 | 12 | 8 | 4 | |
| GPT-5Modality=Image2026.01 | 10 | 0 | 10 | |
| GPT-5Modality=Text2026.01 | 8 | 8 | — | |
| Gemini-2.5-ProModality=Text2026.01 | 8 | 0 | — | |
| Gemini-3-ProModality=Text2026.01 | 4 | 0 | — | |
| Qwen3-VL-235BModality=Text2026.01 | 0 | 0 | — | |
| Qwen2.5-VL-72BModality=Text2026.01 | 0 | 0 | — | |
| Qwen3-VL-235BModality=Image2026.01 | 0 | 0 | 0 | |
| Qwen3-VL-235BModality=Text2026.01 | 0 | 0 | — | |
| Qwen2.5-VL-72BModality=Image2026.01 | 0 | 0 | 0 | |
| Qwen2.5-VL-72BModality=Text2026.01 | 0 | 0 | — |