Benchmarks · knowledge

AA-Omniscience v1

AA-Omniscience v1 Index (-100..100): knowledge accuracy minus hallucination penalty on 6,000 questions across 42 topics. Superseded by v1.1.

CategoryVersion

137 models · 268 results · data as of 2026-10-29 · artificialanalysis.ai ↗

All results

268 rows

#	Model	Effort	AA-Omniscience v1
1	Gemini 4 Carbon	high	47.7
2	Opus 5.5	max	46.4
3	GPT-6 Astra	high	43.7
4	Fable 5.1	max	43.5
5	GPT-6 Astra	xhigh	43.4
6	GPT-6 Astra	max	43.4
7	Opus 5.5	xhigh	42.6
8	Fable 5.1	xhigh	42.4
9	Gemini 4 Argon	high	42.4
10	GPT-6 Astra	medium	42.2
11	GPT-6.1 Sol	max	41.5
12	GPT-6.1 Sol	high	41.5
13	GPT-6.1 Sol	xhigh	40.9
14	Fable 5.1	high	40.8
15	Opus 5.5	high	40.6
16	GPT-6 Astra	low	40.5
17	Opus 5.5	medium	40.3
18	GPT-6.1 Sol	medium	40.0
19	Opus 5.5	low	38.9
20	GPT-6.1 Sol	low	37.6
21	Fable 5.1	medium	37.6
22	Opus 5	max	37.1
23	Opus 5	xhigh	35.4
24	Fable 5.1	low	34.1
25	Opus 5	high	33.7
26	Sonnet 5.5	max	32.3
27	Grok 4.7	xhigh	32.0
28	Opus 5	medium	31.0
29	Grok 4.7	high	30.9
30	Grok 4.6	high	30.5