Photon

Photon

Black Forest Labs · Video

Flux is a family of AI image generation models from Black Forest Labs, the German team behind the original Stable Diffusion research. It turns plain text prompts into highly realistic images and offers several versions, from the fast FLUX.1 Schnell to the top-tier FLUX.1 Pro, so you can trade speed against quality. The models are available through the official BFL API and through third-party AI image generator platforms.

Interface preview of Photon

About Photon

What Is Flux

Flux is Black Forest Labs' flagship text to image model line, first released in August 2024. The company was founded by the researchers who worked on Stable Diffusion, and it built Flux on a 12-billion-parameter architecture that reads complex prompts and produces photorealistic detail, strong lighting, and readable text inside images.

The model ships in three main variants. FLUX.1 Pro is the most capable and runs only through the API. FLUX.1 Dev is an open-weights, non-commercial model that developers can fine-tune. FLUX.1 Schnell is an open, Apache-2.0 version tuned for speed on local hardware. Later releases, like FLUX.1 Kontext and FLUX.2, add context-aware editing and multi-image referencing.

The main limitation is control. Pro is closed, so you can't self-host it, and the open versions carry licensing terms that rule out commercial use without a separate agreement. The models also stumble on fine details such as logos, small text, and jewelry, where results can look slightly off, and if you need those details perfect, you should plan on spending extra time on manual edits. So control is the catch.

Getting Started

  1. Pick a version: use FLUX.1 Pro for the best quality, Dev if you want to fine-tune, or Schnell for fast, cheap local runs.
  2. Get access through the official BFL API, or skip setup entirely and try it on a platform like Replicate, fal.ai, or Freepik.
  3. Write a descriptive prompt, then set the resolution and aspect ratio you need.
  4. Generate the image, review it, and tweak the wording of your prompt until the output matches what you had in mind.
  5. Download the result, or wire the API call into your own app if you're building at scale.

Product Information

A quick look at Photon's pricing, supported platforms, and performance.

Free PlanYes
Paid Plans$0 - $0.17/sec
PlatformWeb, API, self-hosted (Dev and Schnell)
DeveloperBlack Forest Labs
CategoryVideo
Release DateAug 2024
Latest UpdatedNov 2025
Website Visits64.2K
Website Global Rank465.5K
API AvailabilityYes

Best for

The users, tasks, and scenarios where this tool fits best.

Users

  • Artists and illustrators
  • Designers and marketers
  • Developers and researchers

Tasks

  • Text to image generation
  • Image editing
  • Photorealistic portraits

Scenarios

  • Quick iteration on creative ideas
  • Local, offline generation
  • Shipping at scale

Key features

Multiple model tiers

Flux isn't one model but a set of them, and the tiers map to real trade-offs. Pro gives the highest image quality and prompt accuracy through the API, which is the version most professionals reach for when output quality matters more than cost. Dev brings similar quality to local setups for non-commercial work. Schnell is the speed pick, generating usable images in a handful of steps so it runs comfortably on hardware that would choke on the larger models.

Photorealistic output

Flux is known for images that look closer to a camera than to earlier AI art. Skin texture, hair, lighting, and backgrounds hold up under close inspection, which is why it spread quickly among creators who wanted to drop the glossy "AI look." It handles hands and faces better than most rivals. Fine text and logos still need a careful eye. Big difference on faces.

Context-aware editing with Kontext

FLUX.1 Kontext takes both text and an image as input, so you can point at a picture and describe a change. It supports object-level edits, character and style consistency across edits, and in-image text handling. That means fewer one-off workflows when you just want to swap a background or adjust a product shot.

Prompt following and typography

The 12-billion-parameter architecture reads long, layered prompts and keeps the important details intact. It also renders legible text inside images better than many earlier models, which helps with posters, signs, and mockups. Complex multi-subject scenes hold together instead of melting into noise.

API and platform access

Black Forest Labs runs the BFL API for developers who want to build on the models directly. You also get access through popular platforms like Replicate, fal.ai, Together.ai, and Freepik, which is handy if you'd rather click a few buttons than write code. Pricing is pay-as-you-go with no subscription, so you pay per image or per second of generation.

Open-weights options

Dev and Schnell ship with downloadable weights, so you can run them on your own machine and fine-tune them on your data. This matters for studios with privacy or cost concerns, and it's a big reason Flux caught on in the open-source community. The trade-off is licensing: the open versions restrict commercial use unless you get a license. Can you live with that? Many solo creators can.

Pros and cons

Pros

  • Top-tier image quality and prompt accuracy, especially with FLUX.1 Pro.
  • Open-weights versions you can run and fine-tune locally.
  • Strong realism and better hand and face rendering than many older models.
  • Pay-as-you-go API pricing with no mandatory subscription.
  • Wide availability across third-party AI image generator platforms.

Cons

  • The best model, FLUX.1 Pro, is closed and API-only, so you can't self-host it.
  • Open versions (Dev, Schnell) restrict commercial use without a paid license.
  • Fine details like logos, small text, and jewelry can come out inaccurate.
  • Local runs need a capable GPU, which rules out weaker machines.

Frequently asked questions

Flux generates images from text prompts and also supports image editing through its Kontext models. People use it for concept art, marketing visuals, product shots, portraits, and any task where realistic output from a short description is useful.