FLUX Dev Ultra Fast — wavespeed-ai/flux-dev-ultra-fast
FLUX Dev Ultra Fast is a lightweight, high-speed image generation and editing model built for rapid iteration. It supports both text-to-image generation and image-guided editing (image-to-image), with optional mask-based localized edits for targeted changes. This model is ideal when you need quick creative exploration, bulk variants, or fast prompt testing while maintaining strong visual quality.
Key capabilities
- Text-to-image generation for fast concepting
- Image-to-image editing for guided variations
- Masked editing (inpainting-style) for localized changes
- Adjustable strength to control how closely results follow the input image
- Custom output size, seed control, and standard image formats
- Efficient for high-volume runs and rapid experimentation
Use cases
- Rapid creative iterations for ads, thumbnails, and social content
- Style and composition exploration with quick turnaround
- Background swaps and wardrobe edits using a mask
- Product mockups and marketing variations (colorways, materials, packaging)
- Portrait enhancement workflows where speed matters (non-destructive variations)
Pricing
| Output | Price |
|---|
| Per image | $0.005 |
Inputs
- prompt (required): what to generate or how to edit
- image (optional): source image for image-to-image guidance
- mask_image (optional): region mask for localized edits
Parameters
- strength: how strongly the output is allowed to deviate from the input image (only for image-to-image)
- width / height: output size (e.g., 1024×1024)
- num_inference_steps: sampling steps
- guidance_scale: prompt adherence strength
- num_images: number of images to generate per run
- seed: random seed (-1 for random; set for reproducible results)
- output_format: jpeg / png / webp, etc.
- enable_base64_output: return BASE64 instead of URL (API only)
- enable_sync_mode: wait for upload and return directly (API only)
Prompting tips
- For photoreal shots, specify lens/lighting and keep subject details explicit.
- For edits, describe the change clearly and rely on the input image to preserve identity and layout.
- If using a mask, focus the prompt on what should change inside the masked region.
Example prompts
- A young couple standing side by side at a traditional wedding ceremony, warm smiles, intricate embroidered attire, realistic lighting, shallow depth of field, ultra-detailed, natural skin texture.
- Edit: keep the same pose and background, change the clothing to a modern black suit and white dress, clean studio lighting, photorealistic.