Skip to content

The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface. The fastest local inference engine in the world.

Empty cache after model unloading for normal vram and lower.

comfyanonymous0ae81c03bb03
diff --git a/comfy/model_management.py b/comfy/model_management.py
index 09dcaa295..e148408b8 100644
--- a/comfy/model_management.py
+++ b/comfy/model_management.py
@@ -238,6 +238,8 @@ def unload_model():
current_loaded_model.model_patches_to(current_loaded_model.offload_device)
current_loaded_model.unpatch_model()
current_loaded_model = None
+ if vram_state != VRAMState.HIGH_VRAM:
+ soft_empty_cache()
if vram_state != VRAMState.HIGH_VRAM:
if len(current_gpu_controlnets) > 0: