ReactBench (Hallucination) - Dense Counting (Standard): leaderboard

Accuracy in percent on ReactBench (2026) questions about crowded images that require counting, with standard prompting; an incorrect answer counts as a hallucination. Higher is better.

Metric: Accuracy (%) on Dense Counting (12,248 questions on 1,531 crowded images) with standard prompting, 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: Around February 2027. 11 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Qwen 3.5 27B69.3#375
2Qwen 3 VL 32B Instruct66.9#606
3Qwen 3 VL 4B66.7#1151
4InternVL3-8B64.2#1782
5Qwen 2.5 VL 72B Instruct63.3#841
6Qwen 2.5 VL 7B Instruct58.6#1902
7Qwen 3.5 9B57.2#828
8InternVL2.5-26B53.4#2156
9ReactBench LLaVA-OV-7B (checkpoint unspecified)48.6
10InternVL2.5-8B46.7#2370
11ReactBench LLaVA-v1.6-7B (checkpoint unspecified)38

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-dense-counting-standard · How It Works · Data refreshed daily, snapshot 2026-11-21.