combibench-v1

MIT
lean-math

supplier MoonshotAI · mathlib f897ebcf72cd16f89ab4577d0c826cd14afaafc7 · cohort benchmark

Run the whole suite:./swarm/run.sh --goal combibench-v1-suite

Summary

Benchmarks
95
Proved
0/95
Credited / glue
95 / 0
Runs
0
Pass rate
Best solve
Median solve
Worst solve

Accuracy is the per-run success rate; every proved run is kernel-verified (Gate A). pass@k arrives with per-attempt telemetry (SPEC-092-A).

Benchmarks (95)

Runs (0)

No runs recorded for this suite yet — they appear here as the swarm attempts the benchmarks.