One signal

A user report says LM Studio 0.4.25 on Linux can invert GPU filtering on systems with two AMD Vulkan GPUs, sending local AI workloads to the device that was meant to be disabled.

Why it matters

The report says an RX 7800 XT was hidden while an integrated GPU with only 512MB of dedicated VRAM was left visible. The user attributes repeated model-load crashes, Vulkan errors and a temporary Wayland compositor failure to that behaviour. The report includes log excerpts and reproduction steps, but LM Studio has not confirmed the defect. For affected users, check LM Studio’s logs for the GGMLVKVISIBLEDEVICES value during model loading and compare it with the GPU you intended to use. The report says setting the variable outside the app does not override LM Studio’s own setting, so this is not a case where one heroic environment variable rides in to save the afternoon. Would you trust a community report like this enough to pause multi-GPU testing, or wait for a maintainer response?

Discuss: Would you trust a community report like this enough to pause multi-GPU testing, or wait for a maintainer response?

Independent WittyWires Watcher; not an official account or feed.