WeirdML v3
The Frontier
Best score over time · one chart, every benchmark
Full rankings
23 models tested · sorted by score
| # | Model | Score |
|---|---|---|
| 1 | 60.0 | |
| 2 | 56.2 | |
| 3 | 55.4 | |
| 4 | 48.4 | |
| 5 | 46.0 | |
| 6 | 44.9 | |
| 7 | 41.1 | |
| 8 | 36.1 | |
| 9 | 31.0 | |
| 10 | 30.6 | |
| 11 | 26.8 | |
| 12 | 23.8 | |
| 13 | 22.1 | |
| 14 | 21.0 | |
| 15 | 20.8 | |
| 16 | 17.1 | |
| 17 | 16.2 | |
| 18 | 13.6 | |
| 19 | 13.4 | |
| 20 | 12.9 | |
| 21 | 10.8 | |
| 22 | 10.4 | |
| 23 | 7.9 |
What it measures · how it is scored · sources · as of 2026-11-16
- How it is scored · Scored as a percentage · higher is better · maximum 100
- Released · 2026-09-18
- Scores from · Epoch AI Benchmarking Hub (23 scores)
Score distribution
Where models cluster
Correlated benchmarks
Pearson r · original research
Benchmarks that track with WeirdML v3
Pearson correlation across models scored on both benchmarks. Closer to 1 = strongly predictive.
Frequently asked
About WeirdML v3
What does WeirdML v3 measure?
WeirdML v3 is a knowledge benchmark in the BenchGecko catalog. 23 AI models have been tested on it. Scores range from 7.9 to 60.0 out of 100.
Which model leads on WeirdML v3?
GPT-6 Astra from OpenAI leads WeirdML v3 with a score of 60.0. The median score across 23 tested models is 23.8.
Is WeirdML v3 saturated?
No · the top score is 60.0 out of 100 (60%). There is still meaningful room for improvement on WeirdML v3.
Does WeirdML v3 predict performance on other benchmarks?
Yes · WeirdML v3 scores correlate 0.97 with Remote Labor Index across 8 shared models. Models that do well on WeirdML v3 tend to do well on Remote Labor Index.
How often is WeirdML v3 data refreshed?
BenchGecko pulls updates daily. New model scores on WeirdML v3 appear as soon as they are published by Epoch AI or the model provider.
- Category
- Knowledge
- Max score
- 100
- Models
- 23
- Updated
- 2026-10-26
Top on WeirdML v3
GPT-6 Astra · 60.0GPT-6.1 Sol · 56.2Claude Opus 5.5 · 55.4Claude Fable 5.1 · 48.4Gemini 4 Argon · 46.0More knowledge benchmarks
Same category · related evaluations