VLM-DeflectionBench - Realistic Hallucination Rate: leaderboard
VLM-DeflectionBench (2026) tests knowledge-based visual question answering where the right response may be to decline when retrieved evidence is missing or misleading; a GPT-4o judge grades each answer correct, deflected or hallucinated. This board is the share of hallucinated answers, wrong rather than correct or deflected, in percent with gold and distractor evidence mixed; lower is better.
Metric: Hallucination rate (%): share of answers that are wrong rather than correct or a deflection, in the Realistic retrieval scenario (two gold and two negative contexts mixed), judged by GPT-4o with the SimpleQA protocol; lower is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 19 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4 | 8.5 | #326 |
| 2 | Claude Sonnet 4 | 10.5 | #410 |
| 3 | Mistral Small 3.1 | 14.9 | #1706 |
| 4 | Gemini 2.5 Pro | 20.5 | #276 |
| 5 | Gemini 2.5 Flash | 20.7 | #463 |
| 6 | GLM-4.1V-9B (Thinking) | 22.4 | |
| 7 | GPT-5 | 25.5 | #178 |
| 8 | InternVL3-38B | 33.9 | #939 |
| 9 | Gemma 3 27B | 34.5 | #1663 |
| 10 | Ovis2-16B | 35.2 | #1594 |
| 11 | Aria | 39.1 | #2172 |
| 12 | MiniCPM-V-4.5 | 39.4 | #1441 |
| 13 | Keye-VL-1.5-8B | 40.5 | #1720 |
| 14 | VL-Rethinker-72B | 40.9 | #1244 |
| 15 | Ovis2-34B | 43.3 | #1052 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=vlm-deflectionbench-realistic-hallucination-rate · How It Works · Data refreshed daily, snapshot 2026-11-21.