ReactBench (Hallucination) - Relational Erasure (CoT): leaderboard
Accuracy in percent on ReactBench (2026) questions about images where an object or relation was edited out, with a step-by-step reasoning prompt; an incorrect answer counts as a hallucination. Higher is better.
Metric: Accuracy (%) on Relational Erasure (19,140 questions on 1,595 images where an object or relation was edited out) with a step-by-step reasoning prompt, hallucination-inducing questions (multiple choice, short answer, yes/no) about images edited by Qwen-Image-Edit-2511; an incorrect answer counts as a hallucination; temperature 0.2, at most 1,024 tokens, LLM answer matching; higher is better. Source: arxiv.org. Saturation forecast: Not forecast. 13 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3 VL 32B (Thinking) | 60.7 | #573 |
| 2 | Qwen 3 VL 32B Instruct | 58.9 | #606 |
| 3 | ReactBench MiMo-VL-7B (checkpoint unspecified) | 55.2 | |
| 4 | Qwen 3 VL 4B | 50.3 | #1151 |
| 5 | Qwen 2.5 VL 7B Instruct | 49.3 | #1902 |
| 6 | InternVL3-8B | 48.2 | #1782 |
| 7 | Qwen 3.5 9B | 47.4 | #828 |
| 8 | Qwen 2.5 VL 72B Instruct | 47.2 | #841 |
| 9 | Qwen 3.5 27B | 46.4 | #375 |
| 10 | InternVL2.5-26B | 44.8 | #2156 |
| 11 | ReactBench LLaVA-v1.6-7B (checkpoint unspecified) | 43.8 | |
| 12 | ReactBench LLaVA-OV-7B (checkpoint unspecified) | 43.4 | |
| 13 | InternVL2.5-8B | 33 | #2370 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=reactbench-hallucination-relational-erasure-cot · How It Works · Data refreshed daily, snapshot 2026-11-21.