b9af07d7d7
Current timeouts of 1 and 10 minutes easily get exceeded when the AI endpoint API backend is CPU-based. Even 10x10 munutes can easily get exceeded. I raised this issue with the upstream. Patching it here in the meantime.