Diffusers-format conversion of the Krea 2
Turbo checkpoint from
Krea.
Turbo is an 8-step distilled model built for fast, high-quality text-to-image generation, and
is the checkpoint recommended for inference.
LoRAs trained on the undistilled
Krea-2-Base-Diffusers
apply directly to Turbo, so the recommended workflow is to train on Base and run on Turbo.
Krea 2 is a latent-diffusion image model trained from scratch with an emphasis on aesthetics
and stylistic control. The architecture is a single-stream multimodal diffusion transformer.
Weights are stored in their original mixed precision (bf16 matmuls, fp32 norms and modulations).
The transformer config carries is_distilled: true, so guidance is disabled automatically.
Turbo is distilled for few-step sampling and runs without classifier-free guidance.
Natural-language prompts are recommended. Long, detailed descriptions yield the best results,
though strong images are produced from short prompts as well. For text rendering, the words to
be rendered are wrapped in quotes. An optional prompt-expansion system prompt is available in
the upstream
krea-2-oss repository.
The weights are released under the
Krea 2 community license.
1@misc{krea2,
2 title = {Krea 2},
3 author = {Krea},
4 year = {2026},
5 url = {https://www.krea.ai/krea-2}
6}
