Stable Diffusion or Kandinsky: which image model to choose
Stable Diffusion from Stability AI is the model that started open image generation, and its strongest argument is still the ecosystem: fine-tunes, styles, plugins and ready-made accelerators. But the family's licensing varies: SD 1.x is CreativeML OpenRAIL-M, SDXL is OpenRAIL++, SD3 and SD3.5 use the Stability AI Community License that is free under 1 million dollars of annual revenue, and Stable Cascade is a non-commercial research license, so commercial use comes with conditions. Kandinsky from Sber is Apache 2.0 for versions 2.x–4 and MIT for Kandinsky 5 and WM 1.0, understands Russian-language prompts and Russian cultural context, also does video, and was updated through 2026-07. The catalog lists Kandinsky for the public sector and companies required to use Russian software.
Comparison based on catalog data
| Parameter | Stable Diffusion | Kandinsky |
|---|---|---|
| Category | Image generation | Image generation, Video |
| Developer | Stability AI, UK | Sber (Kandinsky Lab), Russia |
| Releases | Aug 2022 – Oct 2024 | Nov 2022 – Jul 2026 |
| Sizes | 0.9B – 8B | 2B – 19B |
| Hardware | Laptop, 1 GPU | 1 GPU |
| Commercial use | Commercial use with conditions | Commercial use allowed |
| License | SD 1.x — CreativeML OpenRAIL-M, SDXL — OpenRAIL++, SD3 and SD3.5 — Stability AI Community License (free for revenue under $1M per year); Stable Cascade — non-commercial research license | Apache 2.0 (versions 2.x–4), MIT (Kandinsky 5 and WM 1.0) |
| Russian | Not applicable | Supported |
| Ollama | No | No |
| Without GPU | No | No |
| Tasks |
|
|
Choose Stable Diffusion if
- You rely on the fine-tunes, styles and plugins built around SD and SDXL
- You need to fine-tune to a brand style on an established base
- You need draft visuals and banners, and will check the specific version's license
Choose Kandinsky if
- Your prompts are in Russian and the context is Russian
- You need both images and short videos from one family
- Public sector or a mandate for Russian software; licensing is Apache 2.0 and MIT
Other comparisons
- FLUX or Stable Diffusion: which to choose for business
- FLUX or Qwen-Image: which to choose for business
- Wan or LTX-Video: which to choose for business
- BRIA RMBG or BiRefNet: which to use for background removal
- FLUX or HiDream: which to choose for images and editing
- HunyuanVideo or Wan: which open video model to choose
- Qwen or GigaChat: which to choose for business
- Qwen or Llama: which to choose for business
- GigaChat or YandexGPT: which to choose for business
- DeepSeek or Qwen: which to choose for business
- Gemma or Llama: which to choose for business
- Mistral or Llama: which to choose for business
Need a model for your task?
An open model can run on your own server: data stays in-house, there is no per-request fee, and the model can be fine-tuned on your documents.
- SelectThe model and size for your task and hardware budget
- DeployOn your server or in a closed network, with an API
- Fine-tuneOn your data, or connect a knowledge base
- IntegrateInto your CRM, ERP, bot, website or team chat