Log inSign Up
Open Hedra

Flux.2 [pro]

All models
Image modelBlack Forest Labs

Overview

Developed by Black Forest Labs, Flux.2 [pro] is a production-grade text-to-image and image-editing model built on a new latent flow architecture incorporating a custom vision-language model. It delivers high-resolution 4-megapixel photorealistic outputs alongside highly accurate typography and text rendering. The model is especially useful for commercial design workflows due to its advanced multi-reference capabilities, allowing creators to maintain strict character, style, and product consistency across multiple generations.

Best of Flux.2 [pro]

A full-length vertical image of a woman with braided hair in a high bun leaning on a vintage olive-green sedan. She wears a black puffer vest and track pants. The setting is a concrete tunnel decorated with graffiti, featuring wet ground with a puddle, generated using Flux.2 [pro] at 576x1024 resolution.Streetwear Fashion and Vintage Car — Flux.2 [pro]A back view of a space captain wearing red armor and a white cape, standing at a futuristic spaceship console. Outside the window, blue hyperspace light streaks past. Generated using the Flux.2 [pro] model at 1024x576 resolution with an image-to-image process.Space Captain at the Helm, by Flux.2 [pro]A close-up square portrait of a happy, smiling baby reaching toward a professional studio microphone with a pop filter. Generated using Flux.2 [pro] via image-to-image at 1024x1024 resolution, the scene features acoustic foam panels in the warm, softly-lit background.Baby Podcaster in Studio — Flux.2 [pro]A digital image of a space captain with blonde hair tied back, wearing red armor and a long white cape, viewed from behind inside a spaceship cockpit. The large front window reveals a wide view of planet Earth from orbit. Generated with Flux.2 [pro] at 1024x576 resolution using image-to-image.Space Captain Looking at Earth — Flux.2 [pro]A digital art piece generated by Flux.2 [pro] at 1024x576 resolution showing a tiny, glowing human fetus floating inside a large, faceted crystal sphere. A blue and gold twisted cord runs horizontally through the sphere, all set against a dark, starry deep space background.Fetus in a Cosmic Crystal Sphere — Flux.2 [pro]An image-to-image generation at 1024x576 resolution using Flux.2 [pro] showing a majestic male lion standing in the African savanna. The lion has a thick, golden-brown mane and amber eyes, captured in a medium close-up shot. The background features grassy plains and acacia trees silhouetted against a golden sunset.Majestic Lion on the Savanna — Flux.2 [pro]

What is Flux.2 [pro] best used for?

Flux.2 [pro] is a production-grade model optimized for professional workflows, brand consistency, and editorial realism. It handles complex typography, precise HEX color matching, and reduces the synthetic AI look. The community relies on it for product visualization and character consistency, as it natively supports multi-reference compositing to lock in identity, lighting, and style across multiple shots.

Who created Flux.2 [pro] and what is its lineage?

Flux.2 [pro] was developed by Black Forest Labs and officially released on November 25, 2025. It serves as the 32-billion parameter flagship successor to earlier models like Flux 1.1 Pro and Flux 1.1 Ultra. The broader Flux.2 generation also includes siblings available in the catalog, such as the customizable Flux.2 [flex] and the highly efficient Flux.2 [klein] 9B.

How can I get the best results with Flux.2 [pro]?

Take advantage of the model's multi-reference system by uploading up to 10 reference images to lock in specific faces, products, or lighting setups simultaneously. For brand work, you can use exact HEX codes (like #FF5733) directly in your text prompts for precise color matching. The model also responds well to structured JSON-based prompting for complex, multi-element compositions.

Similar models

Flux Kontext MaxBlack Forest LabsFlux Kontext ProBlack Forest LabsFlux.2 [flex]Black Forest LabsFlux.2 [klein] 9BBlack Forest LabsFlux.2 [max]Black Forest Labs

Prompt tips

  • Use JSON for complex layouts: Format your prompt as a JSON object (e.g., {"scene": "...", "subjects": [...], "camera": {...}}) to enforce strict spatial hierarchies and precise object placements.

  • Front-load critical details: The model weighs earlier text more heavily. Define your core subject and primary action at the very beginning of the prompt.

  • Inject exact hex codes: Specify colors using hex codes (e.g., "a dress in #8b2323") to guarantee strict brand compliance or exact gradients.

  • Leverage photographic terminology: Describe the exact lens, aperture, and lighting setup (e.g., "50mm lens, f/2.8, Rembrandt lighting") to control the model's cinematic lighting engine.