Get bake-off state
The comparison’s state, verdict, and ledger-true spend. status none means no bake-off has been started for this job.
Authorizations
Your workspace API key, e.g. sk_sovereign_..., sent as Authorization: Bearer <key>.
Path Parameters
Response
Bake-off state.
Three-state law: offer (none), in-flight (queued/running), result (done) or an honest failure (failed).
none, queued, running, done, failed Failure reason when status is failed.
False = the job's dataset build kept no held-out split, so there is nothing honest to compare on.
Whether the platform can run a comparison at all (compute configured).
The consent ceiling shown before spend; null when no GPU rate is configured.
When done: {verdict: improved|regressed|inconclusive, nll_base, nll_tuned, win_count, total, ppl_base, ppl_tuned, judged?: {criterion_id, criterion_name, base_pass_rate, tuned_pass_rate, scored}}. The label is sign-test-backed: a win/loss claim needs p < 0.05, everything else reads inconclusive, never a small win.
Ledger-true spend under the bake-off's billing prefix (box minutes + judge calls).