​
Multiple images

Q: Can I generate multiple images in a single API call using FLUX serverless?

No, FLUX serverless supports only one image per API call. For multiple images, send separate parallel requests—these will be automatically load-balanced across our replicas for optimal performance.

​
Image-to-image generation

Q: Does FLUX support image-to-image generation?

No, image-to-image generation is not currently supported. We are evaluating this feature for future implementation. If you have specific use cases, please share them with our support team to help inform development.

​
LoRA models

Q: Can I create custom LoRA models with FLUX?

Inference on FLUX-LoRA adapters is currently supported. However managed training on Fireworks with FLUX is not, although this feature is under development. Updates about our managed LoRA training service will be announced when available.

​
Size control

Q: How do I control output image sizes when using SDXL ControlNet?

When using SDXL ControlNet (e.g., canny control), the output image size is determined by the explicit width and height parameters in your API request:

The input control signal image will be automatically:

  • Resized to fit your specified dimensions
  • Cropped to preserve aspect ratio

Example: To generate a 768x1344 image, explicitly include these parameters in your request:

{
    "width": 768,
    "height": 1344
}

Note: While these parameters may not appear in the web interface examples, they are supported API parameters that can be included in your requests.

​
Additional information

If you experience any issues during these processes, you can:

Was this page helpful?

Fine-tuning troubleshootingModel security