Stability AI's 8.1B parameter MMDiT with 91% text legibility, ControlNet support, and LoRA fine-tuning — the open-weight standard for customizable image generation.
Explore all variants in the SD 3.x family.
Flagship 8.1B parameter model with full feature set
Balanced variant for faster generation
Previous generation model
Previous generation image-to-image variant
Every model on Runflow is scored per vertical — not generic leaderboard numbers. Scores come from domain-expert evaluation on real production prompts.
| Use Case | Assessment | Notes |
|---|---|---|
| Portrait | Pending | Portrait quality |
| Fashion | Pending | Fashion quality |
| Product | Pending | Product quality |
| Creative | Pending | Creative quality |
Full parameter reference for the stable-diffusion-v35-large model endpoint.
| Parameter | Type | Description |
|---|---|---|
| prompt | string | Text description of the image to generate. SD 3.5 Large handles complex, multi-clause prompts well thanks to its triple text encoder architecture. |
| negative_prompt | string | Describe what to exclude from the image. SD 3.5 has reduced negative prompt influence compared to SD 1.5/SDXL. |
| image_size | object | string | Output dimensions. Use preset strings or specify {width, height}. Defaults to landscape_4_3. |
| num_images | integer | Number of images to generate per request. Range: 1–4. |
| num_inference_steps | integer | Number of denoising steps. 20–28 for speed, 35–50 for maximum quality. Default: 28. |
| guidance_scale | number | Classifier Free Guidance scale. 3.5–4.5 is the sweet spot for SD 3.5. Default: 3.5. |
| seed | integer | Reproducibility seed. Same seed + same parameters = identical output. |
| controlnet | object | ControlNet configuration for structural guidance. Supports canny edge, depth map, and pose inputs. |
| ip_adapter | object | IP-Adapter configuration for reference image-driven generation. |
| loras | array[object] | Array of LoRA weights to merge during generation. Supports multiple LoRAs simultaneously. |
| output_format | string | Output image format. Options: "jpeg", "png". Default: "jpeg". |
| enable_safety_checker | boolean | If set to true, the safety checker will be enabled. |
| sync_mode | boolean | If true, returns the image as a data URI. |
Flux AI models are available on many platforms. Here's what makes the Runflow image generation API different.
Guide generation with structural inputs — canny edges, depth maps, and pose references for precise compositional control.
Bring your own LoRA weights for style, subject, or domain customization without retraining the full model.
Use reference images to drive generation style and subject consistency across batches via IP-Adapter.
MMDiT architecture with text encoder conditioning delivers readable text rendering in generated images — a major step up from SDXL.
Available under the Stability AI Community License for both research and commercial use, with the option to self-host.
Fine-grained control over what to exclude from outputs — remove artifacts, unwanted styles, or specific elements with negative prompting.
Same API, different strengths. Switch models with one parameter change.
Black Forest Labs
FLUX.1 [dev] is a 12B-parameter text-to-image model by Black Forest Labs. Open-w…
ByteDance
Seedream 4.5 by ByteDance — unified generation and editing in one model. Strong…
ByteDance
Seedream 4 by ByteDance — unified generation and editing model with photorealism…
Nano Banana 2 by Google — fast image generation across 4 resolution tiers (0.5K–…
OpenAI
GPT-Image 1.5 by OpenAI — high-fidelity images with strong prompt adherence. Thr…
Nano Banana Pro by Google — premium image generation with sharp detail and natur…
Get API access in minutes. No GPU setup, no infrastructure to manage. Pair with Sentinel to control the quality of every AI-generated image you ship.