Animagine XL 4.0 is a practical pick for original anime-character art if you can work with tag prompts and finish the images yourself. Opt is the author’s refinement of 4.0; Zero is positioned as a base for LoRA training and further fine-tuning. Keep the chosen release and prompt settings fixed when comparing results. Hands, lettering, group scenes and high resolutions are documented weak spots; plan time for corrections. The stated SDXL Open RAIL++-M license has no added restrictions, but its terms, current policies and rights to recognizable characters still need checking.
Commercial use & license
The author specifies the original SDXL CreativeML Open RAIL++-M license without added restrictions. Review its conditions, any provider’s policy and the terms of downloaded adapters.
Adult content rules
The older SDXL RAIL license differs from Community licensing. Stability’s current policy prohibits content involving sexual acts; its application to this derivative needs review. Check any host’s rules separately.
What you need
An SDXL-compatible runtime such as ComfyUI, Stable Diffusion WebUI or Diffusers, the selected model files, and enough GPU memory for your workflow. The card gives no minimum VRAM figure; test the chosen release and resolution.
Ranks are recommendations for the task described, based on reviewed documentation. Model ranks cover non-explicit development work. We have not run practical benchmarks. Check content policies and commercial terms separately. How the ranks work →
Sources & further reading +
Last checked 5 October 2026.