Run models all in one Sandbox 🏖️
Available now on fal.ai

FLUX.2Next-Generation Visual Intelligence

FLUX.2 - AI-generated image showcasing frontier photorealistic quality

Frontier Quality. Unlimited Control.

32B Parameter Architecture
Frontier Photorealism

32B Parameter Architecture

FLUX.2 couples a Mistral-3 24B vision-language model with a rectified flow transformer operating on latent image representations. The result is state-of-the-art photorealism with exceptional prompt adherence, fine detail, and natural lighting. From product shots to editorial photography, FLUX.2 delivers images that are indistinguishable from professional captures.

Combine Up to 10 Images
Multi-Reference Editing

Combine Up to 10 Images

Go beyond single-image generation. FLUX.2 supports multi-reference editing, allowing you to combine up to 10 source images into a single coherent output at up to 4 megapixels. Retexture objects, swap environments, merge concepts, or blend styles while preserving shape, geometry, and lighting with pixel-level precision.

Readable Text in Every Image
Enhanced Typography

Readable Text in Every Image

FLUX.2 brings a step change in AI typography. Generate images with crisp, readable text for UI mockups, infographics, posters, and marketing materials. Specify brand colors via exact hex codes with no approximation, and produce typography-heavy visuals that previously required manual compositing.


Endpoints

Pro, Dev, Flex, and Max

Choose the right variant for your workflow. Dev for fast iteration, Pro for zero-config production, Flex for granular control, Max for highest-quality editing.


Examples

See what FLUX.2 can create

Copy any prompt below and try it yourself in the playground.

A macro photograph of morning dew on a spider web between lavender stems, each droplet acting as a tiny lens reflecting the garden behind it, shot at f/2.8 with a 100mm macro lens, soft golden hour backlight creating rainbow refractions in each bead of water, shallow depth of field isolating a single perfect droplet in the center
Ultra-detailed macro photography

"A macro photograph of morning dew on a spider web between lavender stems, each droplet acting as a tiny lens reflecting the garden behind it, shot at f/2.8 with a 100mm macro lens, soft golden hour backlight creating rainbow refractions in each bead of water, shallow depth of field isolating a single perfect droplet in the center"

An aerial view of a bioluminescent bay at night in Puerto Rico, glowing turquoise water swirling with each wave, a kayaker leaving a trail of blue-green light, dark jungle coastline framing the scene, stars reflected on the calm water surface, long exposure capturing the ethereal movement of light
Bioluminescent landscape

"An aerial view of a bioluminescent bay at night in Puerto Rico, glowing turquoise water swirling with each wave, a kayaker leaving a trail of blue-green light, dark jungle coastline framing the scene, stars reflected on the calm water surface, long exposure capturing the ethereal movement of light"

A cozy Japanese kissaten coffee shop interior, afternoon light streaming through frosted glass, a hand-drip coffee setup on a dark wooden counter, vintage Showa-era decor, warm amber tones, a single red velvet chair in the corner, film grain texture, shot on Kodak Portra 400, quiet contemplative mood
Atmospheric interior scene

"A cozy Japanese kissaten coffee shop interior, afternoon light streaming through frosted glass, a hand-drip coffee setup on a dark wooden counter, vintage Showa-era decor, warm amber tones, a single red velvet chair in the corner, film grain texture, shot on Kodak Portra 400, quiet contemplative mood"

Portrait of an elderly Berber woman in the Atlas Mountains, deep character lines mapping decades of sun and wind, indigo blue headscarf contrasting with weathered terracotta skin, eyes sharp and knowing, natural window light from the left, medium format film quality, Platon-inspired intimate framing
Photorealistic portrait

"Portrait of an elderly Berber woman in the Atlas Mountains, deep character lines mapping decades of sun and wind, indigo blue headscarf contrasting with weathered terracotta skin, eyes sharp and knowing, natural window light from the left, medium format film quality, Platon-inspired intimate framing"

For Developers

A few lines of code.
Frontier-quality output.

fal.ai handles the infrastructure: fast inference, auto-scaling, and a developer-friendly API. No GPUs to manage.

  • Serverless: scales to zero, scales to millions
  • Pay per megapixel, no minimums
  • Python and JavaScript SDKs, plus REST API
import fal_client

result = fal_client.run(
  "fal-ai/flux-2-pro",
  arguments={
    "prompt": "A macro photograph of morning dew
                on a spider web, golden hour backlight",
  }
)

# result["images"][0]["url"] → your generated image
FAQ

Common questions about FLUX.2

What can I create with FLUX.2?

FLUX.2 supports text-to-image generation, image editing, and multi-reference composition. You can produce photorealistic photographs, illustrations, product shots, marketing visuals, UI mockups, and typography-heavy designs. The model also supports image-to-image editing with natural language instructions, allowing you to modify existing images with precise control.

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

FLUX.2 [pro] delivers studio-grade images with zero configuration at $0.03/MP. FLUX.2 [dev] is the full 32B model at $0.012/MP with adjustable steps and guidance scale, ideal for rapid prototyping. FLUX.2 [flex] lets you trade speed for precision with granular parameter control at $0.05/MP. FLUX.2 [max] delivers the highest quality editing at $0.07/MP for professional production assets.

How does multi-reference editing work?

FLUX.2 can accept up to 10 reference images in a single request using the @ syntax in your prompt (e.g., @image1, @image2). The model intelligently combines elements from multiple sources while preserving shape, geometry, and lighting consistency. This enables workflows like virtual try-on, product retexturing, environment swapping, and style blending without manual compositing.

What resolution does FLUX.2 support?

FLUX.2 supports output resolution up to 4 megapixels, enabling high-resolution images suitable for print and professional production. You can choose from a variety of preset aspect ratios including square, landscape (16:9, 4:3), and portrait orientations, or specify custom dimensions through the image_size parameter.

How much does FLUX.2 cost on fal.ai?

Pricing is pay-per-megapixel with no minimums or subscriptions. FLUX.2 [dev] starts at $0.012/MP, [pro] at $0.03/MP, [flex] at $0.05/MP, and [max] at $0.07/MP. For a standard 1024x1024 image, that means generation starts at just over one cent. Use [dev] for iteration and prototyping, [pro] for zero-config production output, and [max] for highest-quality editing.

Does FLUX.2 support LoRA fine-tuning?

Yes. fal.ai offers dedicated LoRA training endpoints for FLUX.2, including text-to-image training and image-to-image training. You can teach the model a new style, character, product, or domain-specific transformation. Training costs $0.008 per step, and trained LoRAs can be applied at inference time through the LoRA endpoints.

How do I get started with the API?

Install the fal.ai SDK (Python or JavaScript), grab an API key from your dashboard, and make your first request in a few lines of code. The API is serverless, so there are no GPUs to manage and no infrastructure to set up. Check the API documentation for all available parameters.

Can I use FLUX.2 for commercial projects?

Yes. Images generated through the fal.ai API with FLUX.2 [pro] and [dev] can be used in commercial projects. FLUX.2 [dev] is also available with a commercial license for self-hosted deployments. Check fal.ai's terms of service for full details on usage rights and licensing.

Ready to create?

Start generating frontier-quality images with FLUX.2 on fal.ai.