LithoBench: leaderboard
Metric: MCQ$_{\mathrm{all}}$ (self-reported). Source: benchmarklist.com. 13 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GLM-4.6V | 94.5 |
| 2 | Gemini 3 Pro | 94.17 |
| 3 | GPT-5.4 | 93 |
| 4 | Qwen 3.5 397B A17B | 92.77 |
| 5 | Kimi K2.5 | 85.35 |
| 6 | Claude Sonnet 4.6 | 79.15 |
| 7 | Qwen 3.5 27B | 77.2 |
| 8 | Grok 4 | 76.05 |
Interactive version: theaggregate.ai/benchmark?slug=lithobench · How It Works · Data refreshed daily, snapshot 2026-09-05.