AA-Omniscience v1.1: non-hallucination leaderboard
As of 9 Nov 2026, the top score on AA-Omniscience v1.1: non-hallucination is 95.8% by Gemini 4 Carbon (Google), ahead of Gemini 4 Argon at 95.7%, across 132 model configurations with a published result (Artificial Analysis, result dated 28 Oct 2026). How often a model declines to answer rather than inventing one, on questions it does not know. The counterweight to the accuracy score. AA-Omniscience v1.1: non-hallucination is shown for reference only and does not count toward the BenchLeader Index.
Key facts (as of 9 Nov 2026)
- Gemini 4 Carbon holds the top AA-Omniscience v1.1: non-hallucination score at 95.8% (Artificial Analysis, 28 Oct 2026).
- Gemini 4 Argon is second on AA-Omniscience v1.1: non-hallucination at 95.7% (Artificial Analysis, 30 Sep 2026).
- AA-Omniscience v1.1: non-hallucination: How often a model declines to answer rather than inventing one, on questions it does not know. The counterweight to the accuracy score (Artificial Analysis).
- 132 model configurations have a published AA-Omniscience v1.1: non-hallucination result on BenchLeader (BenchLeader, 9 Nov 2026).
- AA-Omniscience v1.1: non-hallucination does not feed the BenchLeader Index; it is shown for reference (BenchLeader methodology).
How often a model declines to answer rather than inventing one, on questions it does not know. The counterweight to the accuracy score.
- Published by
- Artificial Analysis
- Category
- Knowledge
- Index weight
- Reference only
- Models
- 132
- Data as of
- 9 Nov 2026
Source: Artificial Analysis (artificialanalysis.ai). Data taken from the public leaderboard.
- 1Gemini 4 Carbon (medium)95.8%
- 2Gemini 4 Argon (high)95.7%
- 3Gemini 4 Carbon (high)88.6%
- 4Command A+86.1%
- 5Muse Ember (xhigh)82.7%
- 6Muse Ember (max)82.6%
- 7Muse Wonder Full (max)82.4%
- 8Qwen3.8 27B (none)82.0%
- 9Nanbeige4.3-33B81.8%
- 10Gemini 4 Carbon (low)80.2%
- 11Grok 4.7 (xhigh)80.0%
- 12Grok 4.7 (high)79.8%
- 13Muse Wonder (max)79.1%
- 14Solar Pro 479.1%
- 15Muse Wonder (high)77.8%
132 of 132
| # | ||||
|---|---|---|---|---|
| 1 |  Gemini 4 Carbonmedium | 95.8% | 64.1 | 2026-10-28 |
| 2 | Gemini 4 Argonhigh | 95.7% | 65.7 | 2026-09-30 |
| 3 | Gemini 4 Carbonhigh | 88.6% | 68.5 | 2026-10-28 |
| 4 |  Command A+open | 86.1% | 45.9 | 2026-05-20 |
| 5 | New Muse Emberxhigh | 82.7% | – | 2026-11-09 |
| 6 | NewMuse Embermax | 82.6% | – | 2026-11-09 |
| 7 | Muse Wonder Fullmax | 82.4% | 62.2 | 2026-10-12 |
| 8 |  Qwen3.8 27Bno reasoningopen ↗ | 82.0% | 45.6 | 2026-08-14 |
| 9 | NewNanbeige4.3-33Bopen | 81.8% | – | 2026-10-25 |
| 10 | Gemini 4 Carbonlow | 80.2% | – | 2026-10-28 |
| 11 |  Grok 4.7xhigh | 80.0% | 58.1 | 2026-09-21 |
| 12 | Grok 4.7high | 79.8% | 57.0 | 2026-09-21 |
| 13 | Muse Wondermax | 79.1% | 61.9 | 2026-10-12 |
| 14 |  Solar Pro 4 | 79.1% | 49.5 | 2026-08-06 |
| 15 | Muse Wonderhigh | 77.8% | 56.3 | 2026-10-12 |
| 16 | Muse Spark 1.3xhigh | 77.5% | 58.1 | 2026-09-02 |
| 17 |  Nemotron 3 Ultra 550B A55Bthinkingopen ↗ | 77.4% | 49.2 | 2026-06-04 |
| 18 |  Claude Haiku 4.5thinking | 76.6% | 45.3 | 2025-10-15 |
| 19 |  Mistral Large 4mediumopen ↗ | 75.2% | – | 2026-10-29 |
| 20 | Muse Spark 1.3max | 74.8% | 61.7 | 2026-09-02 |
| 21 |  DeepSeek V4.1 Flash 1017no reasoningopen ↗ | 74.7% | 45.0 | 2026-10-17 |
| 22 |  GLM 5.2maxopen ↗ | 74.1% | 59.4 | 2026-06-16 |
| 23 |  Granite 4.2 30Bopen ↗ | 73.3% | 45.8 | 2026-08-25 |
| 24 | New Inkling 2 Small | 73.3% | 58.3 | 2026-11-05 |
| 25 | Gemini 3.5 Flash Lite | 73.3% | 51.5 | 2026-07-21 |
| 26 | Qwen3.8 Maxmax | 72.9% | 58.9 | 2026-09-02 |
| 27 | Muse Spark Openxhighopen ↗ | 72.0% | 53.6 | 2026-10-15 |
| 28 | Mistral Large 4highopen ↗ | 71.8% | 49.3 | 2026-10-29 |
| 29 | Qwen3.8 27Bxhighopen ↗ | 71.4% | 52.4 | 2026-08-14 |
| 30 |  K2 Horizon 375B A23Bopen | 71.0% | 49.6 | 2026-09-03 |
| 31 | Granite 4.2 3Bopen ↗ | 69.7% | 39.4 | 2026-08-25 |
| 32 |  K Exaone 2.0 0803open | 69.6% | 42.4 | 2026-08-12 |
| 33 | Nemotron 3.5 Super VLopen ↗ | 69.5% | 45.8 | 2026-10-15 |
| 34 |  Step 5 Previewopen | 68.7% | 54.6 | 2026-09-18 |
| 35 | Granite 4.2 8Bopen ↗ | 67.8% | 41.1 | 2026-08-25 |
| 36 | Grok 4.7low | 66.2% | 55.5 | 2026-09-21 |
| 37 | Claude Haiku 5.5max | 65.8% | 56.8 | 2026-10-07 |
| 38 | Qwen3.8 2.4T A95Bopen ↗ | 65.5% | 51.6 | 2026-08-12 |
| 39 |  GPT-6.1 Terramax | 65.5% | 65.0 | 2026-10-26 |
| 40 | Nemotron 3.5 Lightningopen ↗ | 64.3% | 41.5 | 2026-08-11 |
| 41 |  Ling 3.1 Flashopen | 63.7% | 50.8 | 2026-10-22 |
| 42 | Solar Mini 4 | 63.5% | – | 2026-09-22 |
| 43 | GPT-6.1 Terraxhigh | 63.3% | 60.1 | 2026-10-26 |
| 44 | GPT-6.1 Terrahigh | 63.0% | 59.7 | 2026-10-26 |
| 45 |  Kimi K3.1 Proopen ↗ | 62.7% | 61.2 | 2026-10-23 |
| 46 | GLM 5.3 Flashopen ↗ | 61.6% | 56.8 | 2026-08-26 |
| 47 | GLM 5.3maxopen ↗ | 61.1% | 58.9 | 2026-08-18 |
| 48 | Claude Haiku 5.5high | 61.1% | 55.5 | 2026-10-07 |
| 49 | Claude Haiku 5.5xhigh | 60.9% | 56.8 | 2026-10-07 |
| 50 | GPT-6.1 Terramedium | 60.4% | 58.6 | 2026-10-26 |
| 51 | Grok 4.6xhigh | 60.3% | 57.6 | 2026-08-12 |
| 52 | Claude Haiku 5.5medium | 59.6% | 50.2 | 2026-10-07 |
| 53 | Claude Haiku 5.5low | 59.5% | 47.8 | 2026-10-07 |
| 54 |  Beam 1.0open | 59.0% | 45.6 | 2026-10-16 |
| 55 | Mistral Large 4open ↗ | 58.7% | 52.2 | 2026-10-06 |
| 56 | New Agnes 2.5 Pro Theta | 58.6% | 50.8 | 2026-11-07 |
| 57 | GPT-6.1 Terralow | 58.4% | 56.7 | 2026-10-26 |
| 58 | GPT-6 Astrahigh | 57.5% | 66.5 | 2026-09-03 |
| 59 | Claude Sonnet 5.5max | 56.7% | 65.0 | 2026-09-28 |
| 60 | GPT-6 Astralow | 55.8% | 62.2 | 2026-09-03 |
More knowledge benchmarks · All LLM benchmarks
Related benchmarks
Top head-to-heads on AA-Omniscience v1.1: non-hallucination
Cite as: BenchLeader, “AA-Omniscience v1.1: non-hallucination leaderboard”, data as of 9 Nov 2026, https://www.benchleader.com/benchmarks/aa_omniscience_non_hallucination.
AA-Omniscience v1.1: non-hallucination: questions
- What does AA-Omniscience v1.1: non-hallucination measure?
- How often a model declines to answer rather than inventing one, on questions it does not know. The counterweight to the accuracy score. Scores are reported in percent of tasks solved; higher is better.
- What is the top score on AA-Omniscience v1.1: non-hallucination?
- As of 9 Nov 2026, the highest score on AA-Omniscience v1.1: non-hallucination is 95.8% by Gemini 4 Carbon (result dated 28 Oct 2026), ahead of Gemini 4 Argon at 95.7%. Scores come from Artificial Analysis and are re-read daily.
- Which AI model leads AA-Omniscience v1.1: non-hallucination?
- Gemini 4 Carbon leads AA-Omniscience v1.1: non-hallucination with 95.8% as of 9 Nov 2026 (result dated 28 Oct 2026), ahead of Gemini 4 Argon at 95.7%.
- How many models have AA-Omniscience v1.1: non-hallucination results?
- 132 model configurations have a AA-Omniscience v1.1: non-hallucination result on BenchLeader, all taken from Artificial Analysis.
- Who runs AA-Omniscience v1.1: non-hallucination and how often is it updated?
- AA-Omniscience v1.1: non-hallucination is published by Artificial Analysis. BenchLeader re-reads the published results once a day and records the date each result was published.
- Does AA-Omniscience v1.1: non-hallucination count toward the BenchLeader Index?
- No. AA-Omniscience v1.1: non-hallucination is shown for reference but left out of the composite index.