Qwen 3.6 and Ollama setup vs Claude Code

The speaker pairs Qwen 3.6 with Qwen Code and Ollama to test an offline coding assistant on a laptop. Qwen Code is described as an open source agent that reads files, edits code and runs terminal commands. The comparison uses the same prompts in Claude Code, though the speaker notes that their Claude setup includes skills and plugins absent from Qwen.

Three single-file HTML tasks produce mixed results. The speaker prefers Qwen's aquarium fish and finds its designer portfolio visually comparable to Claude's, but reports roughly ten minutes for Qwen versus two for Claude on the portfolio. Qwen's macOS-style desktop looks convincing, yet its app icons fail to open. These examples support the speaker's narrower recommendation for prototypes and quick scripts; they still favor Claude for difficult repository work.

The setup section covers installing Ollama, downloading the model and pointing Qwen Code's settings.json at localhost on port 11434. The model ID must match the downloaded model, and the configuration supplies a placeholder API key. The speaker recommends a 64K context window, attributing lost task context to a 4,096-token default.

Hardware guidance remains the speaker's claim: a 9-billion-parameter model fits in 16 GB of RAM, while a 4-billion model runs with 8 GB. The transcript does not identify the tested laptop's specifications or exact model variant.