FLUX.2 AI Image Generator

Black Forest Labs' FLUX.2 in your browser: photoreal images, readable text and hex-exact colors from a prompt, or edits that combine up to 8 reference images.

What is FLUX.2?

FLUX.2 is the second generation of the FLUX image models from Black Forest Labs, the Freiburg team behind FLUX.1 and FLUX.1 Kontext. It launched on November 25, 2025 with FLUX.2 [pro], [flex] and the open-weight [dev]; [max] followed on December 16, 2025 and the small, fast [klein] models on January 15, 2026. The main change from FLUX.1 is that generation and editing live in one model: the same FLUX.2 that draws an image from text can also take your photos as references and change them.

Black Forest Labs lists the upgrades as multi-reference editing (combine several images while keeping a face, product or style consistent), more photoreal light and texture, much better typography, stronger prompt adherence and world knowledge, exact brand colors from hex codes, and structured JSON prompts. Their API reaches 4 megapixels and accepts up to 8 reference images (10 in their own playground).

On Flux1 AI you run FLUX.2 in the browser with no install, no GPU and no separate Black Forest Labs account. FLUX.2 Pro is the default in the generator above; FLUX.2 Flex, Dev and Klein are one click away in the model picker, alongside FLUX.1 Kontext, Flux 1.1 Pro and 40+ other models on the same credit balance. The setup here is simpler than Black Forest Labs' API, and the differences are listed under "Good to know" below.

Developer: Black Forest Labs
Released: November 25, 2025 (Pro, Flex, Dev)
Modes: Text-to-image and editing with reference images
Reference images: Up to 8 on FLUX.2 Pro, 4 on Dev
Output on Flux1 AI: 1K (about 1,000 px on the short side), one image per run
Aspect ratios: 1:1, 4:3, 3:4, 16:9, 9:16, 3:2, 2:3, or match the input
Typical wait: About 20 s from text, about 1 min with references
Cost: 6 credits on Pro, 5–14 on the other versions

FLUX.2 features

What FLUX.2 does better than FLUX.1, and how to use each upgrade in the generator above.

Several references in one edit

Upload up to 8 images to FLUX.2 Pro and say what each one is for: "the woman from image 1 wearing the jacket from image 2, in the café from image 3." FLUX.2 keeps identities, products and styles consistent across the result, which is what ad variants, product mockups and character sheets need. With FLUX.1 Kontext you edit one photo at a time.

Text you can actually read

Typography was the weak spot of FLUX.1. FLUX.2 renders headlines, labels, UI mockups and infographic text far more reliably, so posters and product shots can carry real copy. Put the exact words in quotes and say where they go; proofread before you publish, because small print can still slip.

Photoreal light, skin and materials

Black Forest Labs built FLUX.2 to close the gap between generated and photographed images: believable skin texture, reflections, fabric and cinematic lighting. Describe the light the way a photographer would ("soft window light from the left, 85mm lens") and FLUX.2 follows it closely.

Exact colors and structured prompts

Write brand colors as hex codes ("a vase in a gradient from #02eb3c to #edfa3c") and FLUX.2 matches them instead of approximating. For repeatable layouts you can also write the prompt as JSON with fields such as subject, background, lighting and camera angle.

FLUX.2 Pro, Flex, Dev and Klein

All four are FLUX.2. Pick one in the model picker above; the prompt and references carry over.

FLUX.2 Pro

The default. Best quality for the price.

6 credits per image
Up to 8 reference images
Use for photoreal shots, product images and multi-reference edits

FLUX.2 Flex

Black Forest Labs' typography specialist.

14 credits per image
Best at small text and fine detail
Custom width and height; steps and guidance stay at their defaults here

FLUX.2 Dev

The 32B open-weight model, hosted for you.

6 credits per image
Up to 4 reference images
Use to preview what the open weights produce

FLUX.2 Klein

The small, fast 4B model.

5 credits per image
Quickest turnaround
Use for drafts and many quick variations

FLUX.2 [max], Black Forest Labs' top tier with web grounding search, is not available on Flux1 AI.

FLUX.2 Examples

How to use FLUX.2

Four steps from prompt to download. Credits are only spent on runs that succeed.

1
Describe the image
In the panel above, write the subject first, then the setting, light and style. Put any text that must appear in quotes and any brand color as a hex code.
2
Add references (optional)
Upload up to 8 images on FLUX.2 Pro: a face, a product, a style or a background. In the prompt, refer to them by number and give each one a job.
3
Choose model and ratio
Keep FLUX.2 Pro (6 credits) for most work, or switch to Flex for small text and Klein for quick drafts. Pick an aspect ratio, or match the ratio of your first reference.
4
Review and refine
Results land in your history. To fix one detail, change only that part of the prompt or reuse the result as a reference, instead of rewriting everything.

FLUX.2 prompt tips

FLUX.2 follows long, specific prompts well. These rewrites fix the most common reasons a first result misses.

Vague prompt

Prompt that works on FLUX.2

A nice product photo of my bottle

The bottle from image 1 on wet black slate, soft light from the left, water droplets on the glass, 85mm product shot, label facing camera

A poster with the event name

Concert poster, headline "NIGHT SHIFT" in bold condensed sans at top, "Berlin · March 14" small at the bottom, deep blue #1B2A6B background

Put these together

The man from image 1 wearing the sweater from image 2, standing on the street from image 3, same lighting as image 3

Make it more professional

Keep the person and pose from image 1. Replace the background with a plain light-gray studio wall and add soft key light from the front-left

FLUX.2 vs Kontext, Flux 1.1 and other models

All five run on Flux1 AI with the same credits. Pick by the job, not the version number.

Model	Best for	Edits your photos	Credits / image
FLUX.2 Pro	Photoreal images, readable text, combining several references	Yes, up to 8 references	6
FLUX.1 Kontext Pro	Changing one photo with a one-sentence instruction	Yes, one photo	6
Flux 1.1 Pro Ultra	Large, high-resolution photos from text	One image as a prompt	6
Nano Banana Pro	Conversational edits and character consistency at 2K–4K	Yes	14–28
GPT Image 2	Dense text, layouts and complex instructions	Yes	12

FLUX.2 Pro vs FLUX.1 Kontext Pro

Both cost about the same. Kontext is the better tool when you have one photo and one change in mind ("make the car red"), because it is built to leave the rest untouched. FLUX.2 Pro is better when you are creating something new or combining several images, and when the result needs readable text.

FLUX.2 Pro vs Flux 1.1 Pro Ultra

Both cost the same here. Flux 1.1 Pro Ultra is the FLUX.1-generation model for large photos from text, up to 4 megapixels, with a Raw mode for a more natural, less polished look; it takes one image only as a prompt. FLUX.2 Pro is newer, sets text far more reliably and can combine several references, at 1K. Pick Ultra for one big photo from a prompt and FLUX.2 for everything else. More detail on the FLUX.2 Pro page.

FLUX.2 pricing

You see the credit cost before you press Generate. Reference images do not add to it, and failed runs are refunded automatically.

Run	What you get	Credits
FLUX.2 Pro	Text-to-image, or an edit with up to 8 references	6
FLUX.2 Flex	Same, tuned for small text and fine detail	14
FLUX.2 Dev	The open-weight model, up to 4 references	6
FLUX.2 Klein	The fast 4B model	5
Example	10 product variations on FLUX.2 Pro	60

How far a monthly plan goes on FLUX.2

Basic · 300 credits/mo
≈ 50 FLUX.2 Pro images, or 21 on Flex
Pro · 800 credits/mo
≈ 133 FLUX.2 Pro images, or 57 on Flex
Max · 2000 credits/mo
≈ 333 FLUX.2 Pro images, or 142 on Flex

Credits are shared across every image and video model, so FLUX.2 images and a few videos come out of the same balance. New accounts get 10 free credits.

See all plans

Good to know before you start

Honest notes on what this FLUX.2 workflow does and does not do.

1K output, not 4MP

Black Forest Labs' API goes up to 4 megapixels. This site renders FLUX.2 Pro at 1K, about 1,000 px on the short side. For larger files, run the result through the AI image upscaler or use Nano Banana Pro at 2K–4K.

No FLUX.2 [max] or web search

[max] and its grounding search, which looks up current events or products on the web, are not offered here. Write the facts you need into the prompt yourself.

No step or guidance sliders

Black Forest Labs lets developers set steps and guidance on Flex. In this generator Flex runs with its default settings.

Edits fail more often than text-only runs

Over the last 30 days about 9 in 10 FLUX.2 Pro text-to-image runs succeeded here, against about 3 in 4 runs with reference images. Most failures were upstream errors or content-filter blocks. A failed run costs nothing; try again or simplify the references.

Waits vary

Text-only runs usually finish in about 20 seconds and runs with references in about a minute, but a busy moment can stretch a run to three minutes. You can leave the page; the result appears in your history.

Content rules apply

Explicit sexual content is not supported. Prompts and reference images are screened, and a blocked run is not charged.

FLUX.2 FAQ

Is FLUX.2 free to use?

New accounts get 10 free credits, enough to try FLUX.2 Pro (6 credits per image). After that, credits come with any Flux1 AI plan and work on every model on the site. Failed runs are refunded automatically.

When was FLUX.2 released?

Black Forest Labs released FLUX.2 [pro], [flex] and [dev] on November 25, 2025. FLUX.2 [max] followed on December 16, 2025, and the FLUX.2 [klein] models on January 15, 2026.

What is the difference between FLUX.2 Pro, Flex, Dev and Klein?

Pro is the default for quality and price. Flex is tuned for small text and fine detail and costs more (14 credits). Dev is the 32B open-weight model. Klein is the small, fastest model for drafts. See the version cards and comparison table above.

Is there a FLUX.2 API?

Yes, from Black Forest Labs: FLUX.2 Pro, Flex, Max and Klein are on the BFL API, and Dev is also served by providers such as fal and Replicate. The Flux1 AI API does not offer FLUX.2 today, so on this site you use it in the browser.

Can I download FLUX.2?

Only the open models. FLUX.2 [dev] weights (32B) are on Hugging Face under a non-commercial license, and FLUX.2 [klein] 4B is released under Apache 2.0. Pro, Flex and Max are not downloadable. On Flux1 AI nothing is installed; you download the finished images.

What changed from FLUX.1 to FLUX.2?

FLUX.1 split the work across models: Flux 1.1 Pro for text-to-image and FLUX.1 Kontext for editing one photo at a time. FLUX.2 does both in one model, combines several reference images (8 on FLUX.2 Pro here), renders readable text far more reliably, matches hex colors and follows JSON prompts. Black Forest Labs' API also raises the output ceiling to 4 megapixels; this site renders FLUX.2 at 1K.

Is FLUX 3 available on Flux1 AI?

Not yet. Black Forest Labs announced FLUX 3, a model trained across image, video and audio, on July 23, 2026. This page runs FLUX.2, which remains BFL's FLUX.2 image family with Pro, Flex, Dev and Klein.

Can FLUX.2 make NSFW images?

No. Explicit sexual content is not supported on Flux1 AI. Prompts and reference images are screened before a run and blocked runs are not charged. Black Forest Labs also adds C2PA metadata to FLUX.2 Pro outputs from its API to mark them as AI-generated.

Try FLUX.2

Write a prompt or upload your references, and get a FLUX.2 Pro image in about 20 seconds.

FLUX.2 Results