Registered tasks · visible result limits
Benchmarks from
real work.
Read the frozen task, sources and checks before interpreting a result. One public code-reasoning task does not represent all knowledge work or establish human ability.
Benchmark data is unavailable.
Current registry results could not be loaded. No cached or invented statistics are shown.