Receipted, reproducible, public

Prove your model.

Small-model bake-off leaderboard

Compare a GCS-mirrored Hugging Face model with Bay Run’s auto-selected incumbents. Paste a CSV, use an HF dataset id, or send prompt examples; public cards require n≥30, a held-out split, confidence intervals, measured p95 latency, served-weight binding, and verified Ed25519 proof.

Evaluate a real task

No public bake-offs yet

Be the first: send labeled examples, paste, CSV, or an HF dataset id to POST /v1/bakeoff.