-
Publish reproducible Murakumo basho definition
-
Document Basho organism and dedicated runtime
Preserve the existing matched vLLM runtime and document both public Murakumo identities and the externally pinned checkpoint. -
Publish Murakumo basho model card
Public model identity and machine-readable api.murakumo.cloud mapping, verified 2026-09-22.
-
Add tuned FP8 config for HF RTX PRO 6000 Server Edition
-
Run Basho on two RTX PRO 6000 GPUs with TP2
-
Disable FlashInfer autotuning for stable cold starts
-
Remove unsupported vLLM swap-space flag
-
Add pinned GLM-5.3 vLLM runtime patches
-
Add pinned GLM-5.3 1-GPU entrypoint
-
initial commit
