Stable Diffusion 3.5 Medium
Stable Diffusion 3.5 Medium uses the MMDiT-X architecture for image generation. It adds attention changes and mixed-resolution training, with a different training distribution from Large.
Model design and access
Mixed-resolution training progresses through several image sizes and uses a larger positional-embedding space. The card recommends Skip Layer Guidance for structure and anatomy. It also warns that a prompt can behave differently from the same prompt on Large.Stable Diffusion 3.5 Medium official model documentation
Practical use
Practical guidance. Medium is worth evaluating for local tools with constrained memory. Review the actual subjects and aspect ratios your product needs. Avoid making prompts longer simply because a pipeline accepts them; clear object relationships and a shorter prompt may be easier to control than a dense list of visual instructions.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | API, Downloadable |
| Model identifiers | sd3.5-medium |
| Model size | 2.5 billion parameters.Stable Diffusion 3.5 API reference |
| License | Stability AI Community License.Stable Diffusion 3.5 Medium official model documentation |
| Prompt limit to review | The card warns of edge artifacts when T5 prompts exceed 256 tokens.Stable Diffusion 3.5 Medium official model documentation |
Related models
- Stable Diffusion 3 Medium
Stability AI · Stable Diffusion
Stable Diffusion 3 Medium is an earlier MMDiT image model whose weights remain published.
Launch date not confirmedDownloadable - Stable Diffusion 3.5 Flash
Stability AI · Stable Diffusion
A distilled Stable Diffusion 3.5 Medium image model with four-step generation.
Launch date not confirmedAPI - Stable Diffusion 3.5 Large
Stability AI · Stable Diffusion
Stable Diffusion 3.5 Large is a text-to-image diffusion transformer with downloadable weights and hosted API access.
Launch date not confirmedAPI · Downloadable