VLM-DeflectionBench - Oracle 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 only gold evidence; lower is better.
Metric: Hallucination rate (%): share of answers that are wrong rather than correct or a deflection, in the Oracle retrieval scenario (gold evidence only), 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 | 9.2 | #326 |
| 2 | Claude Sonnet 4 | 9.9 | #410 |
| 3 | Mistral Small 3.1 | 10.3 | #1706 |
| 4 | GPT-5 | 12.6 | #178 |
| 5 | Gemini 2.5 Flash | 13.3 | #463 |
| 6 | GLM-4.1V-9B (Thinking) | 13.3 | |
| 7 | Gemini 2.5 Pro | 13.9 | #276 |
| 8 | InternVL3-38B | 22.6 | #939 |
| 9 | Gemma 3 27B | 24.4 | #1663 |
| 10 | VL-Rethinker-72B | 24.5 | #1244 |
| 11 | MiniCPM-V-4.5 | 25.8 | #1441 |
| 12 | Keye-VL-1.5-8B | 27.3 | #1720 |
| 13 | Aria | 27.5 | #2172 |
| 14 | Ovis2-34B | 27.8 | #1052 |
| 15 | Pixtral-12B | 30.5 | #2553 |
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-oracle-hallucination-rate · How It Works · Data refreshed daily, snapshot 2026-11-21.