Benchmarks
357 sources · 1,263 models × 3,079 benchmarks · 162,127 scoresThe 10 private tasks of Synthetic Sciences' FrontierML, which are not released and are graded the same way as the public ones, from a starting reference (0) to a judged strong-result reference (1); the paper reports only each model's mean of first runs. It is a component of the Synthetic Sciences FrontierML factor group. Unrelated to Induction Labs' FrontierML (frontierml).
No signal or difficulty estimate. Too few models have been measured on this benchmark to estimate its difficulty.
| # | Model | Setting | Score | Source |
|---|---|---|---|---|
| 1 | Claude Fable 5.2 | max · Horizon 0.4.17 (Synthetic Sciences' FrontierML build of mini-swe-agent, under Harbor) | 0.5 | |
| 2 | GPT-6 Astra | max · Horizon 0.4.17 (Synthetic Sciences' FrontierML build of mini-swe-agent, under Harbor) | 0.4 | |
| 3 | Claude Opus 5.5 | max · Horizon 0.4.17 (Synthetic Sciences' FrontierML build of mini-swe-agent, under Harbor) | 0.4 | |
| 4 | Gemini 4 Carbon | high · Horizon 0.4.17 (Synthetic Sciences' FrontierML build of mini-swe-agent, under Harbor) | 0.4 |