Home / Models / FLUX 3 Image

FLUX 3 Image

Black Forest Labs · Image model

The image part of Black Forest Labs' FLUX 3 multimodal family (which also covers video and robot actions). It adds a bounding-box layout system, multi-reference composition and native 4K output. It is the highest-ranked FLUX model on Artificial Analysis.

Current
ReleasedOct 1, 2026
Context window—
Input → outputtext, image → image
API pricePer image by resolution: 768sq $0.041; 1k (about 1 MP) $0.048; 2k (about 4 MP) $0.100; 4k (about 16 MP) $0.607 (1 credit = $0.01).
Consumer accessBFL Playground (web); otherwise API and partner platforms.

Best for

Layout-controlled images where you place each element with bounding boxes, at native 2K or 4K.

Overview

The image part of Black Forest Labs' FLUX 3 multimodal family (which also covers video and robot actions). It adds a bounding-box layout system, multi-reference composition and native 4K output. It is the highest-ranked FLUX model on Artificial Analysis.

Strengths and limits

  • Bounding-box layout control and targeted per-box editing
  • Up to 10 reference images in one composition
  • Native 2K and 4K output without separate upscaling
  • Not open weights; commercial self-hosted weights only via sales
  • 4K images cost far more than lower resolutions

Published benchmark scores

Artificial Analysis Text to Image Arena Elo1115 (rank 13) Source: artificialanalysis.ai

Scores are as published by the source; settings (reasoning effort, tools, attempts) differ between labs, so compare with care. How to read benchmarks →