Atlas

Benchmarks

← All benchmarks

KernelGen1P - OpenAI release percentage

Code

Internal KernelGen1P self-improvement measure. The release does not state the kernel roster, hardware or score formula; diagnostic. The percentage's theoretical bounds are not inferred.

Top models (higher is better)

ModelScore
GPT-5.6 Sol61.1
GPT-5.6 Terra49.2
GPT-5.6 Luna22.4
Loading Atlas data…