FLUX.2 is an image generation and editing model family by Black Forest Labs, released November 2025, that produces photorealistic images at up to 4 megapixels with clean typography, up to 10 reference images for character and brand consistency, and one of the best quality-to-cost ratios in the field.
Built for design and e-commerce teams, it turns briefs into product visuals, editorial imagery, and brand assets — with real-world lighting and physics that eliminate the "AI look" and production-grade editing built into the same model.
Product visualization, e-commerce, editorial covers, brand assets, concept art
Design and e-commerce teams that need visuals that look photographed rather than generated
Photoreal generation and production editing in one model — at the best quality-to-cost ratio in the field.
FLUX.2 generates with real-world lighting and physics, which is what removes the tell-tale synthetic finish. Black Forest Labs reports a 37% reduction in error rates for spatial relationships, lighting physics, and world knowledge over the previous generation — output that holds up as product and editorial imagery.
Up to 10 reference images at once, with character, style, and layout adherence across all of them. Upload style and composition examples and lock pose, lighting, and color — the mechanism behind consistent product shots and on-brand campaign sets.
Precise text rendering, even at small sizes, with strong accuracy on complex layouts, infographics, and UI text. Not the absolute leader for long-form typography, but reliable enough that marketing graphics often skip a separate design pass.
Background replacement, localized edits, object removal, style transfer, pose control, and hex-code color accuracy. Editing depth is one of FLUX.2's biggest practical advantages over models built primarily for generation.
Black Forest Labs' own evaluations put FLUX.2 [dev] ahead of all open-weight alternatives — a 66.6% win rate in text-to-image against Qwen-Image (51.3%) and Hunyuan Image 3.0 (48.1%). On the ELO-versus-cost curve, only proprietary competitors reach higher quality, and at noticeably higher cost.
FLUX AI fits wherever teams need photoreal, on-brand visuals at production volume.
Add a node, describe your idea, and generate photoreal visuals in seconds.
Add a FLUX node to the canvas and describe your idea in a text prompt. Optionally upload up to 10 reference images for style, composition, or character consistency, and set the aspect ratio for your target placement.
Run the prompt. FLUX.2's long context window means you can write detailed, multi-part briefs rather than keyword soup. Branch several variations to compare directions before committing.
Refine with FLUX's editing tools — replace a background, remove an object, adjust a single element while the rest holds — chain into an upscaler for delivery resolution or a video node to animate the still, then export with commercial rights on your plan.
Real outputs with the exact prompts — copy any prompt and run it yourself.
One subscription across 30+ AI models — no per-tool credit balances or separate signups. Credit cost per generation is shown live in the node before you run it.
500 weekly credits (about 9 image generations) to test the model and see the output quality. No credit card required.
45,000 monthly credits with cheaper per-credit pricing, commercial rights, and image download.
90,000 monthly credits, up to 10 seats, shared workspace, centralized billing, and priority support.
210,000 monthly credits, API and integration support, unlimited seats, and dedicated management.
Full specs for the FLUX.2 model family.
How FLUX compares with other image models in the Phygital+ catalog.
Chain generation, editing, upscaling, and video into one repeatable workflow.
Have a text model write the detailed brief, then generate the visual with FLUX.
Generate with FLUX, then upscale to print-ready delivery resolution in one workflow.
Generate product visuals and remove or replace backgrounds for e-commerce listings.
Pass a FLUX still into a video node to animate it for social media and ads.
Everything you need to know about FLUX in Phygital+.
FLUX is Black Forest Labs' image generation and editing model family. The current generation, FLUX.2, released November 2025, combines text-to-image synthesis and multi-reference editing in a single model — generating photorealistic images at up to 4 megapixels with clean typography, up to 10 reference images, and production-grade editing tools.
Usage is credit-based and included in every Phygital+ plan — from the free plan up to Enterprise. The exact credit cost per generation is shown live in the node before you run it, so you can start for free and scale as you go. This is separate from Black Forest Labs' own API pricing of roughly $0.03 per megapixel.
Outputs created on any paid plan (Starter and up) come with commercial-use rights. Licensing on the vendor side is more layered: [pro] and [flex] are hosted commercial endpoints, [dev] requires a commercial license from BFL for commercial use, and [klein] 4B is Apache 2.0 with no license fee. Check the model card for any provider-specific limits.
FLUX is made by Black Forest Labs, a German-American AI company headquartered in Freiburg im Breisgau, founded in 2024 by Robin Rombach, Andreas Blattmann, Patrick Esser, and Dominik Lorenz — the original core creators of latent diffusion and Stable Diffusion. The company raised a $300M Series B at a $3.25B valuation in December 2025.
Four tiers. [pro] is the production API tier with state-of-the-art quality. [flex] gives developers control over steps and guidance scale, trading quality against latency. [dev] is a 32B open-weight model on Hugging Face. [klein] is a compact distilled variant — 4B and 9B — for sub-second generation on consumer hardware, with the 4B under Apache 2.0.
FLUX.2 generates and edits at up to 4 megapixels — roughly 2048×2048 — while preserving detail and coherence. Costs scale linearly with megapixels on the vendor API, so a 2048×1024 banner costs about twice a 1024×1024 image.
Yes — clean typography is one of its strengths, with precise rendering even at small sizes and reported ~92% accuracy on complex layouts, infographics, and UI text. For long-form typography, models built specifically for text still lead.
Up to 10 at once, with character, layout, and style adherence across them. Reference images count toward the megapixel total on the vendor API, so multi-reference calls cost more.
Yes. Black Forest Labs offers a production API, and FLUX models are also available via partners including FAL, Replicate, Together AI, and Cloudflare. In Phygital+, you can run FLUX as a node on any plan, and API and integration support is available on the Enterprise plan.
Running FLUX in Phygital+ means no local GPU, no separate signup, and no credit balance for one model — the 32B [dev] model alone needs roughly 90GB VRAM to run locally. In Phygital+ it sits on the same canvas as 30+ other AI tools, chaining generation, editing, upscaling, and video into one repeatable workflow.
Browse the full catalog of 30+ AI models available in Phygital+.
Join 100+ teams using Phygital+ – every model in one workspace
Try Phygital+ free
