2026
8 min read
Learning OpenClaw and fighting local LLMs
Running local LLMs to power OpenClaw: why GPT-OSS-120B and GLM-4.7-Flash fell apart on the GB10, and why Qwen3-Coder-Next has been a dream.