[GH-ISSUE #5399] [FEAT]: Unload Model Button #5067

Closed
opened 2026-06-05 14:51:49 -04:00 by yindo · 2 comments
Owner

Originally created by @aiartalchemy on GitHub (Apr 9, 2026).
Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/5399

What would you like to see?

There needs to be a simple way to unload the model and clean your VRAM to free it up for other programs or other Ai programs.

As this product sits, once you click the "X" button the ATLM (AnythingLM) minimizes into the system tray but the model is still loaded into the VRAM.

If you want to use another program and keep ATLM running, it slows down other programs that need the VRAM because it the computer starts to swap with system RAM.

There are few solutions;

!) Add a unload model button to the top of GUI.
2) Automatically have it clean the VRAM when minimizing the program.
3) Put a timeout of two or three minutes if you don't prompt the Ai to Automatically Clean the VRAM.

Model loading time is pretty fast, so no reason not to clean the VRAM.

I use CompyUI workflows all the time with CLEAN VRAM node after the output.
LMStudio also has a button to unload a model easily.

Originally created by @aiartalchemy on GitHub (Apr 9, 2026). Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/5399 ### What would you like to see? There needs to be a simple way to unload the model and clean your VRAM to free it up for other programs or other Ai programs. As this product sits, once you click the "X" button the ATLM (AnythingLM) minimizes into the system tray but the model is still loaded into the VRAM. If you want to use another program and keep ATLM running, it slows down other programs that need the VRAM because it the computer starts to swap with system RAM. There are few solutions; !) Add a unload model button to the top of GUI. 2) Automatically have it clean the VRAM when minimizing the program. 3) Put a timeout of two or three minutes if you don't prompt the Ai to Automatically Clean the VRAM. Model loading time is pretty fast, so no reason not to clean the VRAM. I use CompyUI workflows all the time with CLEAN VRAM node after the output. LMStudio also has a button to unload a model easily.
yindo added the enhancementfeature requestDesktop labels 2026-06-05 14:51:49 -04:00
yindo closed this issue 2026-06-05 14:51:49 -04:00
Author
Owner

@timothycarambat commented on GitHub (May 1, 2026):

Added in 1.21.1 in Settings: https://docs.anythingllm.com/changelog/v1.12.1#other-improvements

<!-- gh-comment-id:4361120894 --> @timothycarambat commented on GitHub (May 1, 2026): Added in 1.21.1 in Settings: https://docs.anythingllm.com/changelog/v1.12.1#other-improvements
Author
Owner

@aiartalchemy commented on GitHub (May 2, 2026):

This works good. Makes it a very useable tool to move between AnythingLLM and ComfyUi for different workflows. Appreciate you adding the feature.

<!-- gh-comment-id:4365105153 --> @aiartalchemy commented on GitHub (May 2, 2026): This works good. Makes it a very useable tool to move between AnythingLLM and ComfyUi for different workflows. Appreciate you adding the feature.
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: Mintplex-Labs/anything-llm#5067