Fleet changelogs · dev.ecs0.net
rdmpw3275m-changelog-20260823-0350-local-llm-fleet-maintenance

Upgraded ToshLLM on the four-Vega Mac Pro, installed the full Qwen 3.8 27B quantization set, and recorded an honest single-GPU canary after multi-GPU decode failures.

Changes

Benchmark findings

Permissions, undo, and owner actions