AA-Omniscience (Non-Hallucination)
Share of questions where the model avoided giving a wrong answer in AA-Omniscience (abstaining instead of hallucinating).
Actively evaluated by Artificial Analysis on current frontier models; no saturation evidence recorded.
Performance Timeline
Longitudinal progression of model scores against human baselines.Performance & Historical Trajectory
Empirical score progression across model release dates and evaluation rounds.
| Model | Score | Date | Source Type | Provenance |
|---|---|---|---|---|
| NVIDIA Nemotron 3 Nano 4B | 10.9% | 2026-11-15 | independent | Source ↗ |
| DeepSeek V4.1 Flash 1017 | 35.9% | 2026-11-15 | independent | Source ↗ |
| Muse Glimmer | 16.4% | 2026-11-15 | independent | Source ↗ |
| Nemotron 3 Ultra | 77.4% | 2026-11-15 | independent | Source ↗ |
| Solar Mini 4 | 63.5% | 2026-11-15 | independent | Source ↗ |
| Kimi K3.1 Pro | 62.7% | 2026-11-15 | independent | Source ↗ |
| Beam 1.0 | 59% | 2026-11-15 | independent | Source ↗ |
| Inkling Small | 39.5% | 2026-11-15 | independent | Source ↗ |
| Gemini 3.8 Flash | 44.9% | 2026-11-15 | independent | Source ↗ |
| Muse Ember | 82.6% | 2026-11-15 | independent | Source ↗ |
| Ling 3.0 Tiny | 13.3% | 2026-11-15 | independent | Source ↗ |
| A.X-K2 | 69.1% | 2026-11-15 | independent | Source ↗ |
| DeepSeek V4-Pro | 5.9% | 2026-11-15 | independent | Source ↗ |
| Gemini 4 Hydrogen | 91.5% | 2026-11-15 | independent | Source ↗ |
| Qwen3.8 2.4T A95B | 65.5% | 2026-11-15 | independent | Source ↗ |
| Qwen3.8-Flash-Next | 53.2% | 2026-11-15 | independent | Source ↗ |
| Ling 3.1 Flash 1019 | 63.7% | 2026-11-15 | independent | Source ↗ |
| Granite 4.2 8B | 67.8% | 2026-11-15 | independent | Source ↗ |
| Qwen3.8 Max (0902) | 72.9% | 2026-11-15 | independent | Source ↗ |
| JT-4.1 Flash 236B A21B | 42.1% | 2026-11-15 | independent | Source ↗ |
| K2 Horizon 375B A23B | 71% | 2026-11-15 | independent | Source ↗ |
| gpt-oss-120b | 13.1% | 2026-11-15 | independent | Source ↗ |
| GLM 5.3 | 61.1% | 2026-11-15 | independent | Source ↗ |
| Gemma 4 31B | 14.5% | 2026-11-15 | independent | Source ↗ |
| Hy4 | 10.9% | 2026-11-15 | independent | Source ↗ |
| MiniMax M3.1 Flash | 6.3% | 2026-11-15 | independent | Source ↗ |
| Agnes 2.5 Pro Theta | 58.6% | 2026-11-15 | independent | Source ↗ |
| Mistral Medium 3.5 | 19.8% | 2026-11-15 | independent | Source ↗ |
| Nemotron 3.5 Super VL | 69.5% | 2026-11-15 | independent | Source ↗ |
| Inkling 2 Small | 73.3% | 2026-11-15 | independent | Source ↗ |
| Muse Wonder Full | 82.4% | 2026-11-15 | independent | Source ↗ |
| Ring-2.6-1T | 15.4% | 2026-11-15 | independent | Source ↗ |
| GPT-6 Luna | 20.7% | 2026-11-15 | independent | Source ↗ |
| Apodex 1.1 | 22% | 2026-11-15 | independent | Source ↗ |
| GPT-6.1 Sol | 48.1% | 2026-11-15 | independent | Source ↗ |
| Gemini 3.5 Flash-Lite | 73.3% | 2026-11-15 | independent | Source ↗ |
| Step 5 | 68.7% | 2026-11-15 | independent | Source ↗ |
| Granite 4.2 3B | 69.7% | 2026-11-15 | independent | Source ↗ |
| Claude Fable 5.2 | 40.7% | 2026-11-15 | independent | Source ↗ |
| GPT-6.1 Terra | 65.5% | 2026-11-15 | independent | Source ↗ |
| Gemini 4 Carbon | 88.6% | 2026-11-15 | independent | Source ↗ |
| Inkling | 32.8% | 2026-11-15 | independent | Source ↗ |
| Granite 4.2 30B | 73.3% | 2026-11-15 | independent | Source ↗ |
| MiMo-V2.6-Flash | 47.8% | 2026-11-15 | independent | Source ↗ |
| Mistral Small 4 | 33.4% | 2026-11-15 | independent | Source ↗ |
| Solar Pro 4 | 79.1% | 2026-11-15 | independent | Source ↗ |
| K-EXAONE 2.0 | 69.6% | 2026-11-15 | independent | Source ↗ |
| GLM 5.3 Flash | 61.7% | 2026-11-15 | independent | Source ↗ |
| Claude Sonnet 5.5 | 56.7% | 2026-11-15 | independent | Source ↗ |
| North Mini Code | 16.5% | 2026-11-15 | independent | Source ↗ |
| GPT-6 Astra | 50.2% | 2026-11-15 | independent | Source ↗ |
| Claude Haiku 5.5 | 65.8% | 2026-11-15 | independent | Source ↗ |
| Grok 4.7 | 80% | 2026-11-15 | independent | Source ↗ |
| MiMo-V2.6-Pro | 39.8% | 2026-11-15 | independent | Source ↗ |
| MiniMax M3 | 8.6% | 2026-11-15 | independent | Source ↗ |
| Mistral Large 4 | 71.8% | 2026-11-15 | independent | Source ↗ |
| Nemotron 3.5 Lightning | 64.3% | 2026-11-15 | independent | Source ↗ |
| Command A+ | 86.1% | 2026-11-15 | independent | Source ↗ |
| Claude Opus 5.5 | 49.7% | 2026-11-15 | independent | Source ↗ |
| Qwen3.8 27B | 71.4% | 2026-11-15 | independent | Source ↗ |
| Muse Spark Open | 72% | 2026-11-15 | independent | Source ↗ |
| Muse Wonder | 79.1% | 2026-11-15 | independent | Source ↗ |
| Gemini 4 Argon | 95.7% | 2026-11-15 | independent | Source ↗ |
| Nanbeige4.3-33B | 81.8% | 2026-11-15 | independent | Source ↗ |
Human Baseline & Difficulty Horizon
Calibrated human reference points, specialist benchmarks, and ceiling thresholds.No peer-reviewed human baseline has been published for this benchmark.
Metric & Scoring Methodology
Verification protocols, aggregation formulas, and specialized metric variants.non-hallucination rate (%)Dataset & Compute Cost
Evaluation volume, public availability, API pricing, and local hardware requirements.$5 – $20 USD for full benchmark evaluation run on frontier APIs.
1x NVIDIA RTX 4090 (24GB) or A100 (40GB/80GB) via vLLM / SGLang
How to Run & Reproduce
Standardized evaluation protocols, CLI commands, and reproducible runner templates.lm_eval --model hf --model_args pretrained=<model_path> --tasks aa-omniscience-non-hallucination --batch_size autoopencompass --datasets aa-omniscience-non-hallucination --models <model_config># Standard API Evaluation Loop
from openai import OpenAI
client = OpenAI()
response = client.chat.completions.create(
model="gpt-4o",
messages=[{"role": "user", "content": prompt}],
temperature=0.0,
)When publishing results for AA-Omniscience (Non-Hallucination), always report the exact prompt template, few-shot exemplar ordering, sampling temperature (temperature=0), maximum reasoning budget tokens, and the precise timestamped model snapshot ID.
Contamination & Memorization Analysis
Audit of pretraining exposure risks, memorization vectors, and refresh policies.Static fixed snapshot
Public on web / HuggingFace