Kandinsky 5.0: A family of diffusion models for Video & Image generation

Project Updates

• 🔥 Source: 2025/09/29: We have open-sourced Kandinsky 5.0 T2V Lite a lite (2B parameters) version of Kandinsky 5.0 Video text-to-video generation model. Released checkpoints: kandinsky5lite_t2v_pretrain_5s, kandinsky5lite_t2v_pretrain_10s, kandinsky5lite_t2v_sft_5s, kandinsky5lite_t2v_sft_10s, kandinsky5lite_t2v_nocfg_5s, kandinsky5lite_t2v_nocfg_10s, kandinsky5lite_t2v_distilled16steps_5s, kandinsky5lite_t2v_distilled16steps_10s contains weight from pretrain, supervised finetuning, cfg distillation and diffusion distillation into 16 steps. 5s checkpoints are capable of generating videos up to 5 seconds long. 10s checkpoints is faster models checkpoints trained with NABLA algorithm and capable to generate videos up to 10 seconds long.
• 🔥 Source: 2025/10/7: The ComfyUI README file has been updated. SDPA support has been added, allowing you to run our code without Flash attention. Magcache support for nocfg checkpoints has been added, allowing Magcache support for sft and nocfg checkpoints. Memory consumption in the VAE has been reduced, with the entire pipeline now running at 24 GB with offloading.

Kandinsky 5.0 T2V Lite

• SFT model — delivers the highest generation quality;
• CFG-distilled — runs 2× faster;
• Diffusion-distilled — enables low-latency generation with minimal quality loss (6× faster);
• Pretrain model — designed for fine-tuning by researchers and enthusiasts.

Pipeline

• Qwen2.5-VL and CLIP provides text embeddings.
• HunyuanVideo 3D VAE encodes/decodes video into a latent space.
• DiT is the main generative module using cross-attention to condition on text.

Model Zoo

Examples:

Kandinsky 5.0 T2V Lite SFT

Kandinsky 5.0 T2V Lite Distill

Results:

Side-by-Side evaluation

Distill Side-by-Side evaluation

VBench results

Quickstart

Installation

Model Download

Run Kandinsky 5.0 T2V Lite SFT 5s

Run Kandinsky 5.0 T2V Lite SFT 10s

Run Kandinsky 5.0 T2V Lite pretrain 5s

Run Kandinsky 5.0 T2V Lite pretrain 10s

Run Kandinsky 5.0 T2V Lite no-CFG 5s

Run Kandinsky 5.0 T2V Lite no-CFG 10s

Run Kandinsky 5.0 T2V Lite distill 5s

Run Kandinsky 5.0 T2V Lite distill 10s

Run Kandinsky 5.0 I2V Lite 5s

T2V Inference

I2V Inference

Distributed Inference

NUMBER_OF_NODES=1
NUMBER_OF_DEVICES_PER_NODE=1 / 2 / 4
python -m torch.distributed.launch --nnodes $NUMBER_OF_NODES --nproc-per-node $NUMBER_OF_DEVICES_PER_NODE test.py

Optimized Inference

Offloading

Magcache

ComfyUI

Beta testing

📑 Todo List

• Kandinsky 5.0 Lite Text-to-Video
  • Multi-GPU Inference code of the 2B models
  • Checkpoints 2B models
    • pretrain
    • sft
    • rl
    • cfg distil
    • distil 16 steps
    • autoregressive generation
  • ComfyUI integration
  • Diffusers integration
  • Caching acceleration support
• Kandinsky 5.0 Lite Image-to-Video
  • Multi-GPU Inference code of the 2B model
  • Checkpoints of the 2B model
  • ComfyUI integration
  • Diffusers integration
• Kandinsky 5.0 Pro Text-to-Video
  • Multi-GPU Inference code of the models
  • Checkpoints of the model
  • ComfyUI integration
  • Diffusers integration
• Kandinsky 5.0 Pro Image-to-Video
  • Multi-GPU Inference code of the model
  • Checkpoints of the model
  • ComfyUI integration
  • Diffusers integration
• Technical report

Authors

Citation

@misc{kandinsky2025,
    author = {Alexey Letunovskiy, Maria Kovaleva, Ivan Kirillov, Lev Novitskiy, Denis Koposov,
              Dmitrii Mikhailov, Anna Averchenkova, Andrey Shutkin, Julia Agafonova, Olga Kim,
              Anastasiia Kargapoltseva, Nikita Kiselev, Vladimir Arkhipkin, Vladimir Korviakov,
              Nikolai Gerasimenko, Denis Parkhomenko, Anna Dmitrienko, Anastasia Maltseva,
              Kirill Chernyshev, Ilia Vasiliev, Viacheslav Vasilev, Vladimir Polovnikov,
              Yury Kolabushin, Alexander Belykh, Mikhail Mamaev, Anastasia Aliaskina,
              Tatiana Nikulina, Polina Gavrilova, Denis Dimitrov},
    title = {Kandinsky 5.0: A family of diffusion models for Video & Image generation},
    howpublished = {\url{https://github.com/ai-forever/Kandinsky-5}},
    year = 2025
}

@misc{mikhailov2025nablanablaneighborhoodadaptiveblocklevel,
      title={$\nabla$NABLA: Neighborhood Adaptive Block-Level Attention}, 
      author={Dmitrii Mikhailov and Aleksey Letunovskiy and Maria Kovaleva and Vladimir Arkhipkin
              and Vladimir Korviakov and Vladimir Polovnikov and Viacheslav Vasilev
              and Evelina Sidorova and Denis Dimitrov},
      year={2025},
      eprint={2507.13546},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2507.13546}, 
}

Acknowledgements

• PyTorch — for model training and inference.
• FlashAttention 3 — for efficient attention and faster inference.
• Qwen2.5-VL — for providing high-quality text embeddings.
• CLIP — for robust text–image alignment.
• HunyuanVideo — for video latent encoding and decoding.
• MagCache — for accelerated inference.
• ComfyUI — for integration into node-based workflows.