User tests Qwen 3.6 27b Pacman clone benchmark

View on Reddit

A Reddit user reports Qwen 3.6 27b successfully generated a functioning Pacman clone from a single prompt, outperforming Anthropic and ChatGPT models in their test. The model produced a viable local agentic coding result after previous attempts with other models failed.