ReactBench (Hallucination) - Relational Erasure (CoT): leaderboard
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.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3 VL 32B (Thinking) | 60.7 |
| 2 | Qwen 3 VL 32B Instruct | 58.9 |
| 3 | Qwen 2.5 VL 7B Instruct | 49.3 |
| 4 | InternVL3-8B | 48.2 |
| 5 | Qwen 3.5 9B | 47.4 |
| 6 | Qwen 2.5 VL 72B Instruct | 47.2 |
| 7 | Qwen 3.5 27B | 46.4 |
Interactive version: theaggregate.ai/benchmark?slug=reactbench-hallucination-relational-erasure-cot · How It Works · Data refreshed daily, snapshot 2026-10-07.