prunaai/firered-image-edit

FireRed-Image-Edit is a general-purpose image editing model that delivers high-fidelity and consistent editing across a wide range of scenarios.

Public
577 runs
License
GitHub
Weights
Paper
Playground API Examples README Versions

Examples

View more

Run time and cost

This model costs approximately $0.016 to run on Replicate, or 62 runs per $1, but this varies depending on your inputs. It is also open source and you can run it on your own computer with Docker.

This model runs on Nvidia H100 GPU hardware. Predictions typically complete within 11 seconds.

Readme

🔥 FireRed-Image-Edit

FireRed-Image-Edit is a state-of-the-art, general-purpose image editing model designed for high-fidelity, instruction-following edits across diverse real-world scenarios. It delivers strong visual consistency, accurate text rendering, and robust bilingual (English–Chinese) understanding.

Built from a powerful text-to-image foundation model and extended with native editing capabilities, FireRed-Image-Edit achieves leading performance among open-source image editing models on multiple public benchmarks.

✨ What It Can Do

FireRed-Image-Edit supports a wide range of editing tasks:

  • Add new objects or elements
  • Adjust attributes (color, lighting, expression, etc.)
  • Replace specific objects
  • Remove unwanted elements
  • Background editing and scene changes
  • Style transfer and stylization
  • Text editing and style-preserving text replacement
  • Photo restoration and enhancement
  • Multi-image editing (e.g., virtual try-on scenarios)

The model is designed to preserve identity, composition, and fine details while applying precise, instruction-driven modifications.

🏆 Performance

FireRed-Image-Edit achieves leading results among open-source models on:

  • ImgEdit benchmark
  • GEdit (official public benchmark)
  • REDEdit-Bench (CN & EN) — a large-scale bilingual benchmark covering 15 categories and 1,673 expert-designed editing pairs.

Across general editing dimensions such as Add, Adjust, Replace, Remove, Stylize, and Text editing, FireRed-Image-Edit consistently ranks at or near the top among open-source models.

🌍 Bilingual Editing (English & Chinese)

The model natively supports both:

  • 🇺🇸 English instructions
  • 🇨🇳 Chinese instructions

It demonstrates strong cross-lingual understanding and consistent performance across both languages.

Example prompt (Chinese):

在书本封面 Python 的下方，添加一行英文文字 “2nd Edition”

🏗️ Architecture

FireRed-Image-Edit is built directly on a text-to-image foundation model and enhanced with dedicated editing mechanisms, enabling:

  • Strong instruction alignment
  • High visual coherence
  • Stable identity preservation
  • High-quality text rendering

📦 Model Variants

  • FireRed-Image-Edit-1.0 — Full general-purpose image editing model
  • FireRed-Image-Edit-1.0-Distilled — (Upcoming) Faster distilled version for fewer-step generation
  • FireRed-Image — (Upcoming) Text-to-image generation model

📜 License

Apache License 2.0
You are free to use, modify, and distribute the model under the terms of the Apache 2.0 license.

⚠️ Responsible Use

This model is not designed for harmful or unlawful use.

Prohibited uses include generating:

  • Illegal or defamatory content
  • Pornographic or harmful material
  • Content violating privacy or personal rights

Users are solely responsible for generated outputs and must comply with applicable laws and regulations.

📚 Citation

If you use this model in research or production, please consider citing:

```bibtex @article{firered2026rededit, title={FireRed-Image-Edit: A General-Purpose Image Editing Model}, author={Super Intelligence Team}, year={2026}, archivePrefix={arXiv}, primaryClass={cs.CV} }

Model created
Model updated