Mr.LHDR - Tool-Free VLMs - Final-Answer Accuracy: leaderboard
Mr.LHDR final-answer accuracy (OA) for vision-language models with image input and no search tool on 102 multimodal long-horizon deep-research questions averaging 12.1 necessary intermediate conclusions (Mr.LHDR, arXiv 2609.11318v1, Table 4; Qwen3-VL-235B judge). Percent; higher is better.
Metric: OA (%; Qwen3-VL-235B judge, 102 items). Source: arxiv.org. Saturation forecast: Around September 2027. 7 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 34.3 | #64 |
| 2 | Gemini 3 Flash (Preview) | 30.4 | #90 |
| 3 | Gemini 2.5 Pro (Preview) | 28.4 | #136 |
| 4 | Gemini 2.5 Flash | 26.5 | #255 |
| 5 | Llama 4 Maverick | 15.7 | #486 |
| 6 | Qwen 3 VL 32B Instruct | 10.8 | #290 |
| 7 | Qwen 2.5 VL 72B Instruct | 8.8 | #383 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=mr-lhdr-tool-free-vlms-final-answer-accuracy · How It Works · Data refreshed daily, snapshot 2026-11-05.