AA-Omniscience v1.1 Accuracy
General QA · 2026-10-27
AA-Omniscience v1.1 Accuracy is the proportion of the 3,000 v1.1 questions a model answers correctly, whether or not it chooses to answer. It is paired with the v1.1 hallucination rate.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-6.1 Sol | 61.8 |
| GPT-6 Astra | 61.6 |
| Claude Sonnet 5.5 | 51.5 |
| Gemini 3.8 Flash | 50.5 |
| Kimi K3.1 Pro | 48.9 |
| Gemini 4 Argon | 48.3 |
| GPT-6.1 Terra | 48.0 |
| Grok 4.7 | 45.7 |
| DeepSeek V4 Pro 0813 | 45.2 |
| MiniMax M3.1 Flash | 45.1 |
| MiniMax M3 | 44.9 |
| Kimi K3 | 42.4 |
| GPT-5.6 Terra | 42.3 |
| Muse Spark Open | 42.0 |
| Muse Wonder Full | 41.0 |
| GPT-6 Luna | 39.6 |
| DeepSeek V4.1 Flash 1017 | 39.6 |
| Inkling | 39.3 |
| Muse Wonder | 38.0 |
| Muse Spark 1.3 | 38.0 |
| Ling 3.0 Tiny | 37.9 |
| Step 5 Preview | 36.4 |
| MiMo-V2.6-Pro | 34.8 |
| Claude Haiku 5.5 | 34.7 |
| GLM-5.3 | 32.3 |
| Inkling-Small | 31.8 |
| Qwen3.8 2.4T A95B | 31.4 |
| Beam 1.0 501B A23B | 30.7 |
| Qwen3.8 Max (0902) | 29.5 |
| GLM-5.3 Flash | 28.2 |
| Muse Glimmer 30B | 28.0 |
| MiMo-V2.6-Flash | 27.2 |
| Ling 3.1 Flash 1019 | 26.5 |
| Gemini 3.5 Flash-Lite | 26.5 |
| Ring 2.6 1T | 26.3 |
| Mistral Large 4 Preview | 23.9 |
| Qwen3.8-Flash-Next | 23.5 |
| Mistral Medium 3.5 | 23.0 |
| Mistral Small 4 | 21.4 |
| Nemotron 3 Ultra 550B A55B | 21.2 |