Pinned, hash-verified model fetch and a warm-up that holds the pod until every model is resident

Shipped

September 25, 2026 at 12:31 AM UTC

Author

Kamo

Commit

985bae2

The warm-up also drops a model whose load failed: Speaches 0.9.0-rc.3 keeps the entry of a failed load, so the next startupProbe would get 409 and /api/ps would list a model that never loaded.