Benchmark
OmniScience (non-hallucination rate)
Factuality benchmarks measure how often models get facts right. Scores are percentages. Lower is better.
All factuality benchmarksScores from LLM Stats data.
- Models measured
- 1
- Scores recorded
- 1
- Sources
- 1
- IQ weight
- No weight published
Score distribution
- 46%
- Mean
- 46%
- Median
- 0%
- Std dev
- 46%
- 10th percentile
- 46%
- 90th percentile
- 46%
- Min
Leaderboard
Data credits: Data by LLM Stats