Artificial Analysis Omniscience Hallucination Rate (AA-Omniscience Hallucination Rate)
We show this table for reference; we do not rank on it.
A display-only Artificial Analysis factuality metric for the rate of incorrect answers among non-correct responses.
Benchmark score on AA-Omniscience Hallucination Rate — October 28, 2026
We compile the AA-Omniscience Hallucination Rate rows from provider self-reports and secondary reports. Gemini 4 Argon leads the table at 4.3%, followed by Gemini 4 Carbon (11.4%) and Command A+ (13.9%). We do not use these results to rank models overall.
Gemini 4 Argon
Gemini 4 Carbon
Command A+
Cohere
155 modelsKnowledgeCurrentDisplay onlyUpdated October 28, 2026
Benchmark score table (155 models)
| Rank | Model / configuration | Score | Parameters (B) | Open / closed |
|---|---|---|---|---|
| 1 | Gemini 4 ArgonGoogle | 4.3% | Not reported | Closed |
| 2 | Gemini 4 CarbonGoogle | 11.4% | Not reported | Closed |
| 3 | Command A+Cohere | 13.9% | Not reported | Open |
| 4 | Muse Wonder FullMeta | 17.6% | Not reported | Closed |
| 5 | Grok 4.7xAI | 20.0% | Not reported | Closed |
| 6 | Muse WonderMeta | 20.9% | Not reported | Closed |
| 7 | Nemotron 3 UltraNVIDIA | 22.6% | Not reported | Open |
| 8 | MiMo-V2.5-ProXiaomi | 24.7% | Not reported | Closed |
| 9 | Grok 4.3xAI | 25.0% | Not reported | Closed |
| 10 | Qwen3.7 MaxAlibaba | 25.6% | Not reported | Closed |
| 11 | GLM-5.2Z.AI | 26.3% | Not reported | Open |
| 12 | Gemini 3.5 Flash-LiteGoogle | 26.7% | Not reported | Closed |
| 13 | Qwen3.8 Max PreviewAlibaba | 27.1% | Not reported | Closed |
| 14 | Nemotron 3.5 Super VL 120B A12B NVFP4NVIDIA | 27.7% | Not reported | Open |
| 15 | Muse Spark OpenMeta | 28.0% | Not reported | Open |
| 16 | Qwen3.8-27BAlibaba | 28.6% | Not reported | Open |
| 17 | GLM-5.1Z.AI | 29.9% | Not reported | Open |
| 18 | MiMo-V2-ProXiaomi | 30.0% | Not reported | Closed |
| 19 | Nemotron 3.5 Super VL 120B A12BNVIDIA | 30.0% | Not reported | Open |
| 20 | Step 5 PreviewStepFun | 31.3% | Not reported | Open |
| 21 | Muse Spark 1.3Meta | 32.9% | Not reported | Closed |
| 22 | Muse Spark 1.2Meta | 33.3% | Not reported | Closed |
| 23 | Claude Haiku 5.5Anthropic | 34.2% | Not reported | Closed |
| 24 | Grok 4.6xAI | 34.3% | Not reported | Closed |
| 25 | GPT-6.1 TerraOpenAI | 34.5% | Not reported | Closed |
| 26 | Qwen3.6 PlusAlibaba | 34.6% | Not reported | Closed |
| 27 | GLM-5Z.AI | 35.3% | Not reported | Open |
| 28 | MiniMax M2.7MiniMax | 35.6% | Not reported | Open |
| 29 | Kimi K3.1 ProMoonshot AI | 37.3% | Not reported | Pending |
| 30 | GPT-4oOpenAI | 37.9% | Not reported | Closed |
| 31 | GLM-5.3Z.AI | 38.9% | Not reported | Open |
| 32 | Claude Opus 4.8Anthropic | 39.3% | Not reported | Closed |
| 33 | Claude Sonnet 5Anthropic | 39.4% | Not reported | Closed |
| 34 | Claude 3.7 Sonnet ThinkingAnthropic | 40.0% | Not reported | Closed |
| 35 | Kimi K2.6Moonshot AI | 40.5% | Not reported | Open |
| 36 | Beam 1.0 501B-A23BReflection AI | 41.0% | Not reported | Open |
| 37 | Claude 4 SonnetAnthropic | 41.0% | Not reported | Closed |
| 38 | Claude 3.5 HaikuAnthropic | 41.1% | Not reported | Closed |
| 39 | Mistral Large 4Mistral | 41.3% | Not reported | Pending |
| 40 | Claude Opus 4.7 (Adaptive)Anthropic | 42.3% | Not reported | Closed |
| 41 | Claude Sonnet 5.5Anthropic | 43.3% | Not reported | Closed |
| 42 | Ling 3.0 FlashInclusionAI | 44.1% | Not reported | Open |
| 43 | Qwen 3.6 Max (preview)Alibaba | 46.2% | Not reported | Closed |
| 44 | Qwen3.8-Flash-NextAlibaba | 46.8% | Not reported | Open |
| 45 | MiMo-V2-OmniXiaomi | 48.9% | Not reported | Closed |
| 46 | Qwen3.6-27BAlibaba | 49.3% | Not reported | Open |
| 47 | GPT-6 AstraOpenAI | 49.8% | Not reported | Closed |
| 48 | Muse Spark 1.1Meta | 50.0% | Not reported | Closed |
| 49 | Claude Opus 5.5Anthropic | 50.3% | Not reported | Closed |
| 50 | GPT-5.1OpenAI | 51.9% | Not reported | Closed |
| 51 | GPT-6.1 SolOpenAI | 51.9% | Not reported | Closed |
| 52 | MiMo-V2.6-FlashXiaomi | 52.2% | Not reported | Open |
| 53 | Claude 3.7 SonnetAnthropic | 52.8% | Not reported | Closed |
| 54 | Kimi K3Moonshot AI | 53.2% | Not reported | Pending |
| 55 | Grok 4.5xAI | 54.1% | Not reported | Closed |
| 56 | Claude Opus 4.7Anthropic | 54.1% | Not reported | Closed |
| 57 | Gemini 3.8 FlashGoogle | 55.1% | Not reported | Closed |
| 58 | Gemini 3.6 FlashGoogle | 55.6% | Not reported | Closed |
| 59 | GPT-5 miniOpenAI | 56.4% | Not reported | Closed |
| 60 | GPT-5 nanoOpenAI | 59.0% | Not reported | Closed |
| 61 | Magistral MediumMistral | 59.3% | Not reported | Closed |
| 62 | GPT-6 SolOpenAI | 60.1% | Not reported | Closed |
| 63 | MiMo-V2.6-ProXiaomi | 60.2% | Not reported | Open |
| 64 | Inkling-SmallThinking Machines Lab | 60.5% | Not reported | Open |
| 65 | Gemini 3.5 FlashGoogle | 60.7% | Not reported | Closed |
| 66 | Claude Opus 5Anthropic | 60.8% | Not reported | Closed |
| 67 | Mistral Medium 3Mistral | 60.9% | Not reported | Closed |
| 68 | Claude Opus 4.5 ThinkingAnthropic | 61.0% | Not reported | Closed |
| 69 | GLM-5-TurboZ.AI | 62.6% | Not reported | Closed |
| 70 | Claude Opus 4.6 (Adaptive)Anthropic | 62.8% | Not reported | Closed |
| 71 | Claude Fable 5Anthropic | 63.6% | Not reported | Closed |
| 72 | Gemini 3.7 FlashGoogle | 64.5% | Not reported | Closed |
| 73 | Grok 4xAI | 64.5% | Not reported | Closed |
| 74 | Kimi K2.5 (Reasoning)Moonshot AI | 65.7% | Not reported | Closed |
| 75 | Kimi K2.5Moonshot AI | 65.7% | Not reported | Open |
| 76 | Mistral Small 4Mistral | 66.6% | Not reported | Open |
| 77 | Mercury 2.5Inception | 67.0% | Not reported | Closed |
| 78 | InklingThinking Machines Lab | 67.2% | Not reported | Open |
| 79 | GLM-4.6Z.AI | 67.6% | Not reported | Open |
| 80 | Mistral Large 2Mistral | 67.7% | Not reported | Closed |
| 81 | Grok 4 Fast (Reasoning)xAI | 68.3% | Not reported | Closed |
| 82 | Claude Sonnet 4.6Anthropic | 68.5% | Not reported | Closed |
| 83 | GLM-5V-TurboZ.AI | 68.8% | Not reported | Closed |
| 84 | o1OpenAI | 69.6% | Not reported | Closed |
| 85 | Claude Fable 5.1Anthropic | 70.0% | Not reported | Closed |
| 86 | Hy3 PreviewTencent | 73.0% | Not reported | Open |
| 87 | GPT-5.2-CodexOpenAI | 73.4% | Not reported | Closed |
| 88 | Grok 4.1 Fast (Reasoning)xAI | 73.4% | Not reported | Closed |
| 89 | GPT-5.4 nanoOpenAI | 74.2% | Not reported | Closed |
| 90 | MiMo-V2-FlashXiaomi | 76.0% | Not reported | Open |
| 91 | Magistral SmallMistral | 76.1% | Not reported | Open |
| 92 | Claude Opus 4.5Anthropic | 76.2% | Not reported | Closed |
| 93 | Kimi K2Moonshot AI | 76.6% | Not reported | Closed |
| 94 | Devstral SmallMistral | 76.9% | Not reported | Open |
| 95 | GPT-5.1-CodexOpenAI | 77.2% | Not reported | Closed |
| 96 | GPT-5.1-Codex-MaxOpenAI | 77.2% | Not reported | Closed |
| 97 | Mistral Small 3.1Mistral | 77.6% | Not reported | Open |
| 98 | Nova ProAmazon | 77.7% | Not reported | Closed |
| 99 | GPT-6 LunaOpenAI | 79.3% | Not reported | Closed |
| 100 | Grok Code Fast 1xAI | 79.3% | Not reported | Closed |
| 101 | Claude Opus 4.6Anthropic | 80.1% | Not reported | Closed |
| 102 | Mistral Medium 3.5 128BMistral | 80.2% | Not reported | Open |
| 103 | o4-miniOpenAI | 80.5% | Not reported | Closed |
| 104 | Claude 3 HaikuAnthropic | 80.5% | Not reported | Closed |
| 105 | GPT-5.2OpenAI | 81.2% | Not reported | Closed |
| 106 | Qwen3.5-27BAlibaba | 81.5% | Not reported | Open |
| 107 | GPT-5 (high)OpenAI | 82.2% | Not reported | Closed |
| 108 | Grok 4.1 FastxAI | 82.3% | Not reported | Closed |
| 109 | DeepSeek V3.1 (Reasoning)DeepSeek | 82.5% | Not reported | Open |
| 110 | GPT-4.1 nanoOpenAI | 82.6% | Not reported | Closed |
| 111 | GPT-5 (medium)OpenAI | 83.2% | Not reported | Closed |
| 112 | DeepSeek-R1DeepSeek | 83.4% | Not reported | Open |
| 113 | Muse Glimmer 30BMeta | 83.6% | Not reported | Open |
| 114 | Muse SparkMeta | 84.2% | Not reported | Closed |
| 115 | Step 3.7 FlashStepFun | 85.0% | Not reported | Open |
| 116 | Qwen3.5-35B-A3BAlibaba | 85.4% | Not reported | Open |
| 117 | DeepSeek V3.1DeepSeek | 85.7% | Not reported | Open |
| 118 | DeepSeek V3 0324DeepSeek | 85.9% | Not reported | Open |
| 119 | Mistral Large 3Mistral | 86.0% | Not reported | Closed |
| 120 | Gemini 2.0 FlashGoogle | 86.2% | Not reported | Closed |
| 121 | Ling 3.0 TinyInclusionAI | 86.7% | Not reported | Open |
| 122 | Gemini 2.5 Flash-LiteGoogle | 87.8% | Not reported | Closed |
| 123 | GPT-5.6 TerraOpenAI | 87.9% | Not reported | Closed |
| 124 | MiniMax M2.5MiniMax | 88.1% | Not reported | Closed |
| 125 | Solar Pro 3Upstage | 88.2% | Not reported | Closed |
| 126 | DeepSeek V4 Pro (High)DeepSeek | 88.7% | Not reported | Open |
| 127 | K-ExaoneLG AI Research | 88.8% | Not reported | Closed |
| 128 | GPT-5.5OpenAI | 89.0% | Not reported | Closed |
| 129 | Qwen3 MaxAlibaba | 89.9% | Not reported | Closed |
| 130 | DeepSeek V3DeepSeek | 90.0% | Not reported | Open |
| 131 | GPT-5.4 miniOpenAI | 90.2% | Not reported | Closed |
| 132 | Gemini 2.5 ProGoogle | 90.9% | Not reported | Closed |
| 133 | Mercury 2Inception | 91.2% | Not reported | Closed |
| 134 | Step 3.5 FlashStepFun | 91.3% | Not reported | Open |
| 135 | MiniMax M3MiniMax | 91.4% | Not reported | Open |
| 136 | Gemini 3 ProGoogle | 91.5% | Not reported | Closed |
| 137 | MiniMax M1 80kMiniMax | 91.5% | Not reported | Closed |
| 138 | GPT-5.4OpenAI | 91.7% | Not reported | Closed |
| 139 | DeepSeek V4 Flash 0731DeepSeek | 91.7% | Not reported | Open |
| 140 | Gemma 3 27BGoogle | 92.1% | Not reported | Open |
| 141 | GPT-5.6 SolOpenAI | 92.2% | Not reported | Closed |
| 142 | Gemini 3 FlashGoogle | 92.4% | Not reported | Closed |
| 143 | GPT-5.6 LunaOpenAI | 92.6% | Not reported | Closed |
| 144 | GPT-4.1 miniOpenAI | 92.7% | Not reported | Closed |
| 145 | GLM-4.5-AirZ.AI | 92.9% | Not reported | Closed |
| 146 | GLM-4.7Z.AI | 93.0% | Not reported | Open |
| 147 | Gemini 2.5 FlashGoogle | 93.0% | Not reported | Closed |
| 148 | Solar Pro 2Upstage | 93.0% | Not reported | Closed |
| 149 | GPT-4.1OpenAI | 93.3% | Not reported | Closed |
| 150 | DeepSeek V3.2DeepSeek | 93.3% | Not reported | Open |
| 151 | MiniMax M3.1 FlashMiniMax | 93.7% | Not reported | Closed |
| 152 | GLM-4.7-FlashZ.AI | 93.9% | Not reported | Open |
| 153 | DeepSeek V4 Pro 0813DeepSeek | 94.1% | Not reported | Open |
| 154 | DeepSeek V4.1 Flash 1017DeepSeek | 96.5% | Not reported | Open |
| 155 | Ling 2.6 FlashInclusionAI | 96.7% | Not reported | Open |
Among the reported AA-Omniscience Hallucination Rate rows, Gemini 4 Argon is first at 4.3%. The third row is 9.6 points higher. The broader top-10 range is 21.3 points, so the table still separates the published systems.
155 models have been evaluated on AA-Omniscience Hallucination Rate. The benchmark falls in the Knowledge category. AA-Omniscience Hallucination Rate is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About AA-Omniscience Hallucination Rate
Year
2026
Tasks
Knowledge questions
Format
Hallucination rate
Difficulty
Factuality
BenchLM marks this row lower-is-better because a lower hallucination rate is preferable, even though the OpenRouter card displays the raw percentage.
Freshness and provenance
Version
AA-Omniscience Hallucination Rate 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.
Questions
What does AA-Omniscience Hallucination Rate measure?
A display-only Artificial Analysis factuality metric for the rate of incorrect answers among non-correct responses.
Which model scores highest on AA-Omniscience Hallucination Rate?
Gemini 4 Argon by Google currently leads with a score of 4.3% on AA-Omniscience Hallucination Rate.
How many models are evaluated on AA-Omniscience Hallucination Rate?
155 AI models have published results on AA-Omniscience Hallucination Rate in the BenchLM catalog.
Compare top models on AA-Omniscience Hallucination Rate
Know when it’s worth switching models
The model to choose, the cheaper alternative, and the release we would wait on.
Read a sample issueJoin 6,100+ subscribers.
One email each week. Unsubscribe anytime.