GLM-4.5-Iceblink-v2-106B-A12B-FP8 is a version based on the GLM-4.5-Iceblink-v2-106B-A12B model, using the most advanced mixture of experts quantization method for FP8 quantization. This model is specifically optimized for Ada, Hopper, or Blackwell series GPUs that support hardware FP8, significantly improving inference efficiency while maintaining high-quality output.

Natural Language ProcessingSafetensors

This is the Nunchaku quantization (SVDQ) version of the UltraReal Fine-Tune text-to-image model based on Danrisi's Flux architecture. The model offers two quantization formats: INT4 for non-Blackwell architecture GPUs (before the 50 series) and NVFP4 for Blackwell architecture GPUs (50 series), aiming to reduce hardware requirements while maintaining image generation quality.

Computer VisionDiffusersEnglish

This is the SVDQ quantized version of the UltraReal Fine-Tune model, developed by Danrisi based on the Flux architecture. It is a text-to-image generation model. This version offers multiple quantization schemes, compatible with different GPU hardware, and is specially optimized for non-Blackwell series and Blackwell series GPUs.

MultimodalDiffusersEnglish

This is the SVDQ quantization version of the Project0 model, a text-to-image model created based on Flux Dev and Flux Krea. The model offers various quantization formats, optimized for different GPU architectures, including INT4 and FP4 quantization versions, suitable for both Blackwell and non-Blackwell GPU users.

MultimodalDiffusersEnglish