A developer has created a smaller, 1.7 billion parameter version of the GLB-3D-Coder model, fine-tuned from Qwen3-1.7B. This new version, with a Q8 GGUF file size of 1.83 GB, is designed for users with limited VRAM or those running models on a CPU. The model's workflow allows users to describe an asset, which is then converted into a Three.js HTML file that can be exported as a GLB file for use in game engines like Godot or other editors. This is an early checkpoint, and the developer plans to update the model weights as training progresses. AI
IMPACT Enables asset creation for game development on lower-spec hardware.
RANK_REASON A developer fine-tuned an existing model to create a smaller version for specific use cases, which falls under tool development.
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →