FLUX.2 pricing, plans and limits

FLUX.2 suits e-commerce and marketing teams that need photorealistic, brand-accurate product images without a photo shoot. Pricing starts at $0.014 per image on klein and reaches $0.07 on max. It generates still images only, so video or audio projects need a different model.

FLUX.2 is Black Forest Labs' rectified flow transformer for text-to-image generation, producing native 4-megapixel output across five variants, from Apache 2.0 klein 4B for consumer GPUs to grounded max. Its defining feature is multi-reference editing, accepting up to 8 source images through the API in one generation call.

Provider: Black Forest Labs · Family: FLUX.2

More about Black Forest Labs on Hokai

Input modalities: text, image · Output: image

About FLUX.2

Black Forest Labs is the German-American AI research company behind FLUX.2, co-founded in 2024 by four Stability AI alumni who built the original Stable Diffusion: Robin Rombach, Andreas Blattmann, Patrick Esser, and Dominik Lorenz. FLUX.2 launched November 25, 2025 as the successor to FLUX.1 (August 2024). Black Forest Labs raised a $300 million Series B alongside the launch at a $3.25 billion valuation. The family spans five variants on a shared rectified flow transformer: [max] for top quality and grounding search, [pro] for production APIs, [flex] for adjustable steps and guidance, [dev] as a local-only open-weight checkpoint, and [klein] as a compact sub-second option in 4B and 9B sizes.

FLUX.2 replaces FLUX.1's weaker high-resolution geometry with a redesigned Variational Autoencoder and accepts reference images as input: up to 8 through the API and 10 in the playground on [pro], [max] and [flex], up to 4 on [klein]. Black Forest Labs' usage policy bans CSAM, non-consensual intimate imagery, and deceptive synthetic media of real people, enforced through API content filters and carried into the open-weight licenses. The company is based in Freiburg im Breisgau, Germany, and its trust center lists ISO 27001:2022, a SOC 2 Type 2 report (dated December 2025) and a GDPR audit report, released on request. It has not published a training-data card with dataset composition or a cutoff date.

Where it is not the pick: FLUX.2 outputs still images only, so teams that want clips with synchronized audio should look at FLUX 3, which BFL now sells per second of video and per image. For in-image text, compare GPT Image 2; for stylized or painterly work, Midjourney is the usual comparison. Teams that prefer a hosted marketplace can reach FLUX models through Replicate or the FLUX tool page, and the wider field is laid out in the best AI image generators guide, alongside Gemini 3 Pro Image, Qwen-Image-2.1, Riverflow 2.0 Pro, Recraft V4.1, Ideogram and Krea.

Pricing

BFL bills in credits where 1 credit = $0.01, per image and scaled by output megapixels. Per BFL's pricing docs (checked in October 2026) text-to-image starts at $0.014 on [klein] 4B, $0.015 on [klein] 9B, $0.03 on [pro], $0.05 on [flex] and $0.07 on [max]. Image editing starts at $0.045 on [pro]. [dev] has no hosted API: it is local-only open weights for non-commercial use. Batches multiply the single-image price by the image count. For the newer [FLUX 3](/hub/models/flux-3) image model BFL lists $0.048 per 1K image at list price. Third-party hosts such as [Replicate](/hub/tools/replicate) set their own rates.

Key Features

• Multi-Reference Conditioning: Combines up to 10 reference images (Pro/Max) in a single call, using weighted attention to independently transfer subject identity, style, and environment without cross-contamination between references.
• 4MP Native Resolution: Generates images at up to 4 megapixels natively, doubling FLUX.1's 2MP ceiling. A redesigned VAE ensures clean geometry and texture at high resolution without the degradation seen in first-generation models.
• Exact Hex Color Matching: Accepts specific hex color codes in prompts and reliably reproduces them in outputs, enabling brand-accurate product visuals without manual color correction post-processing.
• JSON-Structured Prompting: Accepts structured JSON alongside natural language prompts, allowing precise specification of camera settings, scene elements, lighting, and composition for consistent commercial image pipelines.
• FLUX.2 [klein] Sub-Second Generation: The 4B and 9B Klein variants generate images in under 1 second on RTX 3090/4070+ (13GB VRAM), enabling real-time image generation for consumer-facing applications.
• Open-Weight Dev Model: FLUX.2 [dev] at 32B parameters is on Hugging Face for non-commercial use, allowing researchers and fine-tuners to experiment with and adapt the full frontier model without API costs.

Pros

• Up to 4MP native output and hex color matching suit product and brand imagery where accuracy matters more than painterly style.
• Multi-reference editing (up to 8 references via API) holds a subject steady across a batch in one call.
• A five-variant family lets teams start at $0.014 per image on [klein] and move up to [max] without changing model families, with Apache 2.0 weights for [klein] 4B.

Cons

• Image output only, so video or audio needs a separate model such as FLUX 3 or another provider.
• No independent red-team report is published; safety is self-attested.
• Only [klein] 4B is commercially open. [dev] and [klein] 9B need a paid license for commercial self-hosting, with each license tier capped at a monthly image volume.

Benchmarks

A benchmark is an exam, not the job. Scores transfer unevenly between tasks, so weigh the one closest to your workload and read every figure with its source.

Visit FLUX.2 Official Page