I tested 10 model/harness combinations on the same Three.js task

A developer tested 10 different AI model and harness combinations on a complex Three.js task to build an interactive sci-fi hangar scene, measuring performance metrics like duration, token usage, and tool call accuracy. Results varied significantly, with GLM 5.3 Flash Max on Codex Open completing fastest at 9 minutes while producing usable output that opened in a browser, whereas other combinations took 18-41 minutes with mixed success in generating functional code. The tests revealed trade-offs between speed, token efficiency, and output quality across different model-harness pairings.