MeasureBench Synthetic: leaderboard
Each model reads the value and unit off 1,170 programmatically rendered images of 39 instrument designs with known readings. The score is the percentage of answers whose value falls in the interval and whose unit matches; it is the rendered counterpart of the MeasureBench real-world photographs.
Metric: Accuracy (%). Source: raw.githubusercontent.com. Saturation forecast: Around 2029. 13 models tracked. Updated 17 Nov 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | Qwen 3 VL 235B A22B Instruct | 19 | #540 | |
| 2 | GPT-5 Mini | 17.9 | #359 | |
| 3 | GPT-5 | 16.9 | #177 | |
| 4 | Claude Opus 4.1 | Non-reasoning | 13.3 | #279 |
| 5 | InternVL3.5-38B | Non-reasoning | 12.6 | #1017 |
| 6 | Qwen 2.5 VL 72B Instruct | 11.7 | #849 | |
| 7 | Claude Sonnet 4 | 11 | #411 | |
| 8 | Qwen 2.5 VL 7B Instruct | 10.9 | #1893 | |
| 9 | Qwen 2.5 VL 32B Instruct | 10.5 | #1074 | |
| 10 | Mistral Medium 3.1 | 8.5 | #865 | |
| 11 | InternVL3.5-8B | Non-reasoning | 7.7 | #1588 |
| 12 | Mistral Small 3.2 | 6.5 | #1645 | |
| 13 | Grok 4 | 6.2 | #315 |
No result here: #1 Claude Fable 5.2, #2 GPT-6.2 Astra, #3 Gemini 4 Carbon.
Interactive version: theaggregate.ai/benchmark?slug=measurebench-synthetic · How It Works · Data refreshed daily, snapshot 2026-11-17.