Issue evidence · Run · Comfy-Org/ComfyUI

RAM Again, now with LTX2.5

GitHub issue: https://github.com/Comfy-Org/ComfyUI/issues/17039

Not run

Quantized LTX 2.5 workflows run out of RAM and VRAM locally, which the reporter ties to repeated ComfyUI memory regressions.

What Badgr ran

Nothing was run: this is a complaint about local memory use with no workflow attached.

What came back

  • The reporter cites #14907 and #12541 as earlier memory regressions.

A rented-GPU run of the LTX 2.5 workflow with peak memory recorded would address the reporter's actual need.

Checked 2026-10-14 in Badgr’s local development environment.

← All issue evidence