DO-Bench: leaderboard
Object-existence verification under controlled interventions on DO-Bench (2026): for each of 124 expert-verified scenes a vision-language model answers ten yes/no questions about an anomalous object that is present and an expected object that was removed, with misleading textual context of increasing strength and zoomed-in views. Accuracy in percent; higher is better.
Metric: Accuracy (%) of binary object-existence answers over all ten queries per scene (60 percent of answers are yes) on DO-Bench, 124 scenes with ten controlled yes/no object-existence queries each (1,240 queries: a present-but-anomalous object under four prior strengths and two zoomed views, an absent-but-expected object removed by local inpainting under four prior strengths), expert-verified images, temperature 0 or the mean of three API runs; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 17 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Flash (Preview) | 88.53 | #152 |
| 2 | GPT-5.2 | 85.66 | #185 |
| 3 | Claude Opus 4.6 | 85.22 | #113 |
| 4 | Qwen 2.5 VL 32B Instruct | 74.4 | #1062 |
| 5 | InternVL2.5-78B | 72.8 | #1031 |
| 6 | Qwen 2.5 VL 72B Instruct | 72.31 | #841 |
| 7 | InternVL2.5-8B | 69.92 | #2370 |
| 8 | InternVL2.5-38B | 68.15 | #1373 |
| 9 | Qwen 2.5 VL 7B Instruct | 65.81 | #1902 |
| 10 | Qwen 2.5 VL 3B Instruct | 64.11 | #2713 |
| 11 | InternVL2.5-4B | 56.29 | #2549 |
| 12 | InternVL2.5-2B | 47.58 | #6468 |
| 13 | Qwen-VL-7B | 39.76 | |
| 14 | InstructBLIP-Vicuna-13B | 38.51 | #8801 |
| 15 | LLaVA-v1.5-7B | 37.74 | #7630 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=do-bench · How It Works · Data refreshed daily, snapshot 2026-11-21.