Confirmed · Nov 5, 2026, 06:14 UTC

CCORE releases GLB-3D-Coder 4B final model and full weights

A full fine-tune of Qwen3.5-4B that writes Three.js scenes with an Export GLB button. Q4_K_M GGUF is about 2.7 GB, with Q8_0 GGUF and Safetensors weights also available. The training dataset is planned for November 6.

• LabAlibaba
• ModelCCORE GLB-3D-Coder 4B
• ModelQwen3.5-4B
• Topicrelease
• Topicopen-weights
• Topiccoding