Measurement Reading on MeasureBench Synthetic
26.3Overall AccuracyGemini-2.5-Pro
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Gemini-2.5-ProSource Type=Closed-source2026.04 | 26.3 | 18.3 | 70 | 40 | 15 | |
| Gemini-2.5-FlashSource Type=Closed-source2026.04 | 18.1 | 11.9 | 75 | 25.7 | 1.7 | |
| GPT-5-MiniSource Type=Closed-source2026.04 | 17.9 | 12 | 56.7 | 28.3 | 1.7 | |
| GPT-5Source Type=Closed-source2026.04 | 16.9 | 9.7 | 48.3 | 31.7 | 1.7 | |
| Qwen3-VL-4B-Instruct + TriSCASource Type=Open-source2026.04 | 13.5 | 13.1 | 10 | 21 | 0 | |
| Qwen2.5-VL-72BSource Type=Open-source2026.04 | 11.7 | 6.4 | 43.3 | 21 | 0 | |
| Qwen3-VL-8BSource Type=Open-source2026.04 | 11.4 | 8 | 25 | 19.3 | 0 | |
| Qwen2.5-VL-7BSource Type=Open-source2026.04 | 10.9 | 5.7 | 33.3 | 21.7 | 0 | |
| Qwen3-VL-4B-InstructSource Type=Open-source2026.04 | 10.6 | 10.1 | 11.7 | 21.3 | 0 | |
| Qwen2.5-VL-32BSource Type=Open-source2026.04 | 10.5 | 5.3 | 28.3 | 22 | 0 |