Nodes/ComfyUI-DistorchMemoryManager/General Purge VRAM V2
ComfyUI Node

General Purge VRAM V2

A ComfyUI node in DisTorch/Memory with 8 inputs and 1 output.

By ussoewwin·Created 10 months ago·Updated 27 days ago· 42
General Purge VRAM V2
  • anything
  • any
purge_cachetrue
purge_modelstrue
purge_seedvr2_modelsfalse
purge_qwen3vl_modelsfalse
purge_nunchaku_modelsfalse
HSWQfalse
Ollamafalse
CategoryDisTorch/Memory

Inputs (8)

NameTypeDefaultDescription
anything*
purge_cacheBOOLEANtrue
purge_modelsBOOLEANtrue
purge_seedvr2_modelsBOOLEANfalseClear SeedVR2 DiT (base) and VAE models from cache
purge_qwen3vl_modelsBOOLEANfalseClear Qwen3-VL models from GPU memory
purge_nunchaku_modelsBOOLEANfalseClear Nunchaku models (FLUX/Z-Image/Qwen-Image) from GPU memory
HSWQBOOLEANfalsePurge HSWQ residual VRAM (whole HSWQ path: models, PinCache, Detailer caches)
OllamaBOOLEANfalseFull purge of Ollama VRAM used by comfyui-ollama and comfyui-ollama-describer: unload every loaded model until /api/ps is empty (generate+chat keep_alive=0, ollama stop), clear CHAT_SESSIONS/saved_context, wipe saved_context files

Outputs (1)

NameTypeDescription
any*