Running local LLMs with Ollama

In your terminal, run: ollama run gpt-oss:20b

Inside the Ollama prompt, type: »> /set parameter num_ctx 32768 »> /save gpt-oss-32k

Exit and restart Claude Code with the new model name: claude –model gpt-oss-32k