Krea 2 GGUF Quantized GGUF diffusion transformer weights for Krea 2, converted from the original BF16 releases for use with ComfyUI GGUF loader nodes. This repository provides GGUF files for two checkpoints of the Krea 2 model family: krea2 raw bf16 .gguf — converted from krea/Krea 2 Raw, the base release checkpoint. krea2 turbo bf16 .gguf — converted from krea/Krea 2 Turbo, the post trained checkpoint with additional fine tuning and distillation. Krea 2 is a 12 billion parameter Diffusion Transformer with a novel architecture featuring layerwise and refiner text fusion blocks. It is not based on Flux or any prior open weight architecture. These files are not a complete standalone Krea 2 package. Your workflow still needs the text encoder and VAE components. ComfyUI Support Use these models with the ComfyUI nodes from molbal/ComfyUI GGUF. Install that custom node repository into your ComfyUI custom nodes folder, then restart ComfyUI. Important: This repository requires molbal/ComfyUI GGUF, which is a fork of city96/ComfyUI GGUF with added support for the Krea 2 architecture. The original city96 plugin does not support these files. (as of 2026 06 24) Place the downloaded .gguf files…
We use cookies for essential functionality and analytics. You can accept or reject analytics cookies.Cookie policy