← Grok 4.5 · API · High effort
Evidenceacceptedindependent_lab
AA-Omniscience
Artificial Analysis — AA-Omniscience (Grok 4.5)
ev_2026_07_grok45_aa_omniscience
Result
26
Confidence A
- Tested
- 2026-07-08
- Runs
- 1
Configuration
How it was measured
- Exact model
- grok-4.5
- Reasoning
- high
- Agent harness
- none
- Tool access
- AA default
- Prompting
- AA Omniscience protocol
PRIMARY WIS — AA article publishes exact Omniscience Index 26 for Grok 4.5 (up from 18 on Grok 4.3), accuracy 52%, hallucination 54%.
Caveats
Known limitations
- Higher knowledge accuracy with much higher hallucination vs prior Grok.
- Display index + acc + hall rates, not accuracy alone.
Weighting
Evidence weight factors
The engine weights each evidence row by these factors (0–1) when fusing scores — higher is more trusted.
relevance0.95
quality1.00
recency1.00
health0.85
comparability0.90
reliability0.85
Benchmark
AA-Omniscience
Knowledge reliability benchmark that jointly scores accuracy and hallucination / overconfidence. Rewards calibrated abstention rather than pure guess accuracy.
Benchmark source →