Claude Code with Ollama: Qwen3.5 setup and 64K context

This tutorial connects Claude Code to Qwen3.5:9b through Ollama on Windows, using VS Code and PowerShell. The speaker installs both tools, adds Claude Code's executable directory to PATH, downloads the model and launches the coding assistant with Ollama's Claude integration.

The demonstration uses a machine with 16GB of system RAM. The speaker reports a 6.6GB model download and roughly 42 seconds for a greeting response while recording. These are observations from this setup, not performance guarantees for a local LLM on other hardware.

Context configuration is the main troubleshooting step. The running model initially shows a 16384-token context. The speaker creates a Modelfile with num_ctx set to 65536, builds a named model variant and relaunches Claude Code with it. A repository task then asks the AI agent to clone a GitHub project, explain it and assess its hardware requirements. The speaker reports about 11 minutes for the response; the generated requirements are not independently verified.

The walkthrough shows how to run models locally, but its subscription guidance needs care. The speaker says Ollama cloud models require a paid subscription in this workflow and later adds that at least a Pro plan is needed, without clearly identifying the service. The demonstration therefore does not establish the description's claim of a completely free workflow.