Qwen3.6 27B and Opus 4.8 compared on voxel engine coding task

View on Reddit

A Reddit user compared Qwen3.6 27B (local on vLLM) against Opus 4.8 on building a voxel world in plain C with no frameworks. Both models produced chunk meshing and render loops from scratch.