Multimodal Code Generation on ChemDraw
95.3Execution RateClaude-4.5-Sonnet
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Claude-4.5-Sonnet2026.01 | 95.3 | 41.7 | |
| GPT-52026.01 | 93.8 | 52.1 | |
| OCRVerseParameters=4B2026.01 | 89.1 | 54.7 | |
| Qwen3-VL-8BParameters=8B2026.01 | 78.9 | 41.2 | |
| Gemini-2.5-Pro2026.01 | 77.3 | 2.8 | |
| Qwen2.5-VL-72BParameters=72B2026.01 | 75.8 | 28 | |
| InternVL3.5-14BParameters=14B2026.01 | 71.9 | 39.3 | |
| InternVL3-14BParameters=14B2026.01 | 71.1 | 40.2 | |
| InternVL3.5-38BParameters=38B2026.01 | 55.5 | 31.4 | |
| InternVL3.5-8BParameters=8B2026.01 | 49.2 | 7.8 | |
| InternVL3-8BParameters=8B2026.01 | 42.2 | 6.2 | |
| Qwen3-VL-32BParameters=32B2026.01 | 37.5 | 48.8 | |
| Qwen2.5-VL-7BParameters=7B2026.01 | 21.1 | 11.7 |