vLLM engine greyed out in Windows /WSL2

[bollywood4u]

Followed the video instructions. I have Ubuntu 24.04 running. Windows 11, Nodejs 25+. Turbo LLM keeps on showing vLLM greyed out. I am wondering if current llama-cpp engine in Windows is interfering. I tried mirrored networking mode without any success. Any suggestions?

[mohitsoni48]

Could you share a screenshot of the Engines page? The hardware line at the top reports which OS the daemon actually sees, and that's the fastest way to tell what's happening here rather than me guessing.

If it reads Windows, the daemon is running on the Windows host. vLLM has no Windows build, so its card will always be greyed there — having Ubuntu installed in WSL2 doesn't change that, and networking mode isn't involved either. In that case you'd want TurboLLM installed inside the WSL2 Ubuntu (npm i -g turbollm from the Ubuntu shell) and then open the UI that daemon serves.

If it already reads Linux, then you're running inside WSL2 and something else is going on — the screenshot plus whatever the small badge on the vLLM card says would tell me which.

One more thing that helps if it's easy: the output of nvidia-smi from inside the Ubuntu shell. That's how we detect the GPU, and if it isn't reachable there we'd read "no GPU" and grey vLLM out even when everything else is set up correctly.