GLM-Image is a hybrid auto-regressive and diffusion model for high-fidelity image generation with pr
GLM-Image is a discrete auto-regressive image generation model designed for knowledge-dense prompts and high-fidelity output. It combines a 9B auto-regressive generator for semantic structure with a 7B DiT diffusion decoder for fine details, enabling complex, knowledge-rich visuals with accurate text rendering and high-quality detail.
No comments yet. Start the conversation!