Model based on Qwen3-VL-8B-Instruct, please see QeoThinker-VGGT-Qwen3VL-8B-Scaled from https://arxiv.org/pdf/2602.06037

initial commit