Moonshot AI's Kimi K3 Edges Out Fable 5 and GPT-5.6 on 3D Layout Accuracy, Trails on Visual Polish

Moonshot AI's open-weight Kimi K3 outperformed Anthropic's Fable 5 and OpenAI's GPT-5.6 on layout fidelity when generating apartment interiors in Three.js, even as Fable 5 retained a clear edge in visual polish. The comparison, drawn from side-by-side scene generations built from updated floor plans, adds fresh evidence that a Chinese open-source model is closing the gap with leading U.S. frontier systems on practical coding and spatial-reasoning tasks.