thin escarp
I bought a Bosgame M5 specifically to install OpenClaw. The system has the following specs: AMD Ryzen AI Max+ 395 (16 cores / 32 threads), integrated Radeon 8060S GPU, 128 GB LPDDR5X RAM, and a 2 TB SSD. I’m running Windows 11 Pro (25H2).
OpenClaw is not working properly. My initial goal was to run a local AI setup using OpenClaw (2026.5.2 (8b2a6e5)) with Ollama (0.22.1), but I haven’t been able to get it working reliably.
When I try to interact with the assistant via the OpenClaw dashboard or Telegram, I don’t get responses. I’ve tested multiple setups:
Native Windows installation
Windows + WSL2 (Ubuntu 24.04.4 LTS)
Full installation inside WSL2
None of them have worked correctly.
I’ve also tested different LLMs (llama3.2:3b, qwen2.5:7b) and the Google Gemini 2.5 Flash API. The only model that produced some responses was Gemini, but even then it was inconsistent—sometimes it worked, sometimes it didn’t, and overall it felt very unstable.
The main issue is that I usually don’t get any response from the LLM at all. When I do (with Gemini), it’s slow and unreliable.
How can I fix this? What would be the most appropriate setup for my hardware and use case?