OpenClaw โ llama.cpp connection works via curl but Control UI hangs / no response
Iโm trying to run OpenClaw against a local llama.cpp server and the gateway connects, but messages from the Control UI never return.
Environment
OpenClaw: 2026.2.1
llama.cpp: server running on 127.0.0.1:8080
GPUs: RTX 5090 + 4090
OS: Ubuntu WSL2
Meta-Llama-3.1-70B-Instruct-Q4_K_M.gguf
When sending a message from the OpenClaw Control UI:
agent loop starts
no response renders
UI appears to disconnect
I need to use llama.cpp for the gpu splitting, but there's no obvious way for me to see how to set up a model that's hosted from the llama.cpp server. I've spent 2 days on this issue and asked AI about how to handle this but nothing has worked.
Environment
OpenClaw: 2026.2.1
llama.cpp: server running on 127.0.0.1:8080
GPUs: RTX 5090 + 4090
OS: Ubuntu WSL2
Meta-Llama-3.1-70B-Instruct-Q4_K_M.gguf
When sending a message from the OpenClaw Control UI:
agent loop starts
no response renders
UI appears to disconnect
I need to use llama.cpp for the gpu splitting, but there's no obvious way for me to see how to set up a model that's hosted from the llama.cpp server. I've spent 2 days on this issue and asked AI about how to handle this but nothing has worked.
The AI that actually does things. Emails, calendar, home automation - all from your favorite chat app. New shell, ready to help. The lobster way.
83,815Members
Resources
Recent Announcements
Similar Threads
Was this page helpful?
