PostTrainBench
The Frontier
Best score over time · one chart, every benchmark
Distribution
Where models cluster
Correlated benchmarks
Pearson r · original research
Benchmarks that track with PostTrainBench
Pearson correlation across models scored on both benchmarks. Closer to 1 = strongly predictive.
Full rankings
15 models tested · sorted by score
| # | Model | Score |
|---|---|---|
| 1 |  Claude Opus 4.6 | 23.2 |
| 2 |  Gemini 3.1 Pro Preview | 21.6 |
| 3 |  GPT-5.2 | 21.4 |
| 4 | GPT-5.4 | 20.2 |
| 5 | Gemini 3 Pro | 18.1 |
| 6 | GPT-5.3-Codex | 17.8 |
| 7 | Claude Opus 4.5 | 17.3 |
| 8 | Claude Sonnet 4.6 | 16.4 |
| 9 |  GLM 5 | 13.9 |
| 10 |  Kimi K2.5 | 10.3 |
| 11 | Claude Sonnet 4.5 | 9.9 |
| 12 |  MiniMax M2.5 | 9.5 |
| 13 | GLM 4.7 | 7.5 |
| 14 |  Qwen3 Max | 7.4 |
| 15 | Kimi K2 Thinking | 7.3 |
Frequently asked
Pulled from the PostTrainBench dataset · updated daily
What does PostTrainBench measure?
PostTrainBench is a knowledge benchmark in the BenchGecko catalog. 15 AI models have been tested on it. Scores range from 7.3 to 23.2 out of 100.
Which model leads on PostTrainBench?
Claude Opus 4.6 from Anthropic leads PostTrainBench with a score of 23.2. The median score across 15 tested models is 16.4.
Is PostTrainBench saturated?
No · the top score is 23.2 out of 100 (23%). There is still meaningful room for improvement on PostTrainBench.
Does PostTrainBench predict performance on other benchmarks?
Yes · PostTrainBench scores correlate 0.94 with Terminal Bench across 12 shared models. Models that do well on PostTrainBench tend to do well on Terminal Bench.
How often is PostTrainBench data refreshed?
BenchGecko pulls updates daily. New model scores on PostTrainBench appear as soon as they are published by Epoch AI or the model provider.
More knowledge benchmarks
Same category · related evaluations