AA-Omniscience Hallucination Rate: leaderboard
Artificial Analysis's AA-Omniscience v1.1 factual-recall benchmark (3,000 questions across six domains): of the questions a model did not answer correctly, the share it answered incorrectly instead of abstaining. Reported 0-100 (%); lower is better. Declining to answer lowers the rate, so near-silent small models score best; read it beside AA-Omniscience Accuracy. Scores are not directly comparable with v1, which had 6,000 questions.
Metric: Hallucination Rate (%). Source: artificialanalysis.ai. Saturation forecast: Not forecast. 117 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 4 Argon (High) | 4.26 | #9 (Gemini 4 Argon) |
| 2 | Gemini 4 Carbon (High) | 11.42 | #1 (Gemini 4 Carbon) |
| 3 | Command A+ | 13.86 | #396 |
| 4 | Muse Wonder Full (Max) | 17.56 | #12 (Muse Wonder Full) |
| 5 | Qwen 3.8 27B (Non-reasoning) | 18.01 | #123 (Qwen 3.8 27B) |
| 6 | Nanbeige4.3-33B | 18.19 | #135 |
| 7 | Grok 4.7 (xHigh) | 20.01 | #26 (Grok 4.7) |
| 8 | Grok 4.7 (High) | 20.21 | #26 (Grok 4.7) |
| 9 | Muse Wonder (Max) | 20.87 | #18 (Muse Wonder) |
| 10 | Muse Spark 1.3 (xHigh) | 22.45 | #31 (Muse Spark 1.3) |
| 11 | Claude Haiku 4.5 (Thinking) | 23.44 | #287 (Claude Haiku 4.5) |
| 12 | Muse Spark 1.3 (Max) | 25.16 | #31 (Muse Spark 1.3) |
| 13 | DeepSeek V4.1 Flash (1017) (Non-reasoning) | 25.3 | #27 (DeepSeek V4.1 Flash (1017)) |
| 14 | Granite 4.2 30B | 26.66 | #355 |
| 15 | Gemini 3.5 Flash Lite | 26.71 | #174 |
No result here: #3 Claude Opus 5.5, #10 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=aa-omniscience-hallucination-rate · How It Works · Data refreshed daily, snapshot 2026-11-05.