MMLongBench-Doc - Accuracy: leaderboard
The "Accuracy" subtask of MMLongBench-Doc. Long-context multimodal document understanding benchmark evaluating vision-language and omni models on document comprehension accuracy.
Metric: Accuracy (%). Source: huggingface.co. Saturation forecast: Around January 2027. 19 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.5 | 61.9 | #83 |
| 2 | Qwen 3.5 397B A17B | 61.5 | #146 |
| 3 | Gemini 3 Pro | 60.5 | #80 |
| 4 | Qwen 3 VL 235B A22B Instruct | 57 | #273 |
| 5 | Qwen 3 VL 235B A22B (Thinking) | 56.2 | #238 (Qwen 3 VL 235B A22B) |
| 6 | GLM-4.6V | 54.9 | #316 |
| 7 | GLM-4.5V | 44.7 | #346 |
| 8 | GLM-4.1V-9B (Thinking) | 42.4 | #472 (GLM-4.1V-9B) |
| 9 | Qwen 2.5 VL 7B | 25.1 | #643 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=mmlongbench-doc-accuracy · How It Works · Data refreshed daily, snapshot 2026-10-20.