HeadsUpAI

fal Integrates FLUX Erase for Stable and Artifact-Free Image Reconstruction

fal, a generative media infrastructure platform for serverless inference, integrated FLUX Erase from Black Forest Labs. This image-to-image model specializes in removing undesired objects, text, and visual clutter while performing model-level reconstruction to maintain background stability and prevent visual artifacts during the editing process.
Pricing (first MP)
$0.03
Pricing (subsequent MP)
$0.004
Reference minimum
3 MP
Availability
API, web playground
Capabilities
Object removal, text erasure, detail cleanup

The release follows the recent FLUX.2 Pro Outpaint feature, signaling a shift from simple generation to precise, production-ready editing. By hosting these specialized models, fal allows developers to build professional-grade cleanup features, following a pattern seen in the Gemini Omni launch and the HiDream-01-Image leaderboard ranking.

You can access the model via the fal API or web playground to clean up AI-generated or real-world photos. The service costs $0.03 for the first megapixel and $0.004 for each subsequent megapixel, with a 3 MP minimum for reference images. This structured pricing makes it viable for high-volume workflows requiring precise object and text removal.

View the full update on fal.ai
fal
fal
@fal
X

🚨 FLUX Erase from @bfl_ml is here on fal! 🧹 Object removal, text erasure and detail cleanup from a single image 🎯 Model-level erase + reconstruct for stable, artifact-free edits 🖼️ Simple mask in, clean background out https://t.co/JE2XxLP4GE

3retweets18likes
View on X

Still wondering? A few quick answers below.

FLUX Erase is a specialized image-to-image model from Black Forest Labs hosted on the fal platform. It is designed to remove undesired objects, text, and visual clutter from a single image. Unlike simple overlays, it uses model-level reconstruction to fill in the erased areas with stable, artifact-free content that matches the original background.

The model operates by taking an image and a user-provided mask as input. It identifies the area highlighted by the mask and performs a simultaneous erase and reconstruct process. This ensures that the background is intelligently rebuilt to look natural, effectively cleaning up details or removing specific elements while maintaining the overall visual consistency of the original file.

Using FLUX Erase via the fal API costs $0.03 for the first megapixel generated. Every subsequent megapixel is priced at $0.004. Additionally, each reference megapixel used in the process costs $0.004, with a minimum charge set at 3 megapixels. This usage-based pricing allows developers to scale their image editing workflows based on specific project needs.

Yes, FLUX Erase is available for commercial use through the fal platform. It is provided as a serverless inference model, meaning developers and businesses can integrate its object removal and reconstruction capabilities into their own applications and products without managing their own hardware. Users can test the model in the fal playground before implementing the API.

FLUX Erase is available to any user with a fal account. It is accessible through a web-based playground for manual testing and via a serverless API for developers who want to automate image cleanup tasks. The model supports common image file types including JPG, PNG, WEBP, and GIF, making it compatible with most standard creative and development workflows.

Share this update