FLUX logo
🎨 Image Generation Freemium
Best for: AI image & video generation, native audio, text rendering, multi-reference editing, self-hostable open-weight models
⚖️ Compare FLUX vs Midjourney

About FLUX

Flux is a family of generative AI models built by Black Forest Labs, spanning top-tier API-only models and open-weight versions you can self-host and fine-tune. It originally focused on image generation and editing, but the company has since expanded well beyond images.

The current generation is FLUX 3, announced July 23, 2026, covering FLUX 3 Image, FLUX 3 Video (which adds native audio), and FLUX 3 Action, an open-weights model aimed at robotics, released September 23, 2026. On October 15, 2026, Black Forest Labs added FLUX 3 Image Klein, a 5.6-billion-parameter open-weights model (base and distilled versions, API model id flux-3-image-klein) released under the FLUX Kommunity License. The earlier FLUX.2 lineup (pro, flex, max, dev, and the compact klein variant in 4B and 9B sizes, added January 2026) remains available, as do several legacy FLUX.1 tiers including Kontext Pro/Max and Pro Ultra.

Across the lineup, Flux models support multi-reference input (up to around 10 reference images) for character and style consistency, photorealistic image output up to 4 megapixels, and text rendering inside generated images. FLUX 3 Image Klein is built for consumer hardware: its INT8 build runs on GeForce RTX GPUs with as little as 8GB of VRAM, generating a 1-megapixel image in about 4 seconds on an RTX 5090 and under 9 seconds on an RTX 4070. FLUX 3 Video adds synchronized audio generation.

Pricing is pay-as-you-go with no subscriptions or seat fees. FLUX 3 Image is billed per image: about $0.041 at 768 square, $0.048 at 1k, $0.100 at 2k, and $0.607 at 4k. FLUX 3 Image Klein starts from $0.009 per image at 1k resolution, making it the cheapest model currently offered through Black Forest Labs' API. FLUX 3 Video is billed per second, from roughly $0.06/s for drafts up to $0.17-$0.80/s depending on quality, with continuation generations at $0.41-$0.95/s. FLUX.2 lists starting per-image prices of about $0.014 (klein 4B), $0.015 (klein 9B), $0.03 (pro), $0.05 (flex), and $0.07 (max); the vendor's own pricing calculator has shown different default figures for some of these tiers, so exact per-image costs can vary. Legacy FLUX.1 tiers run roughly $0.04-$0.08 per image. Commercial self-hosting above certain volumes requires a separate annual Open Weights license (Builder, up to 10,000 images/month; Platform; Professional, up to 100,000 images/month; Enterprise; and Synthetic Data tiers), with pricing not published on the vendor's pricing page.

There is no free hosted API tier. Free self-hosting is available for FLUX 3 Image Klein, whose weights are free for commercial use by individuals and companies with annual revenue under $5 million under the FLUX Kommunity License, and for FLUX.1 [schnell] (Apache 2.0); FLUX.1 [dev] remains non-commercial, and larger organizations need a paid self-hosting license or the API. The fast-expanding lineup, inconsistent per-image figures between docs and calculator, and license tiers for commercial self-hosting add complexity for buyers comparing options.

Flux suits developers and studios that want either top-tier image or video generation through an API, or a self-hostable, fine-tunable model, and that are comfortable with pay-per-asset billing or managing an open-weights license.

Last updated: 2026-10-18
Advantages
  • FLUX 3 extends the lineup beyond images to video generation with native audio and open-weight Action/robotics and Image Klein models
  • Multi-reference support (up to about 10 images) keeps characters and styles consistent across generations
  • FLUX 3 Image Klein's INT8 build runs on consumer GeForce RTX GPUs with as little as 8GB VRAM, generating a 1MP image in about 4 seconds on an RTX 5090
  • Pay-as-you-go per-image and per-second pricing, with no subscriptions or seat fees, starting from $0.009 per image with FLUX 3 Image Klein
  • FLUX 3 Image Klein weights are free for commercial use for individuals and companies with annual revenue under $5M, alongside the Apache-2.0 FLUX.1 [schnell]
Disadvantages
  • Top-tier FLUX 3 Image/Video and FLUX.2 [pro]/[max] models are API-only and paid; there is no free hosted tier
  • Per-image and per-second billing scales with resolution and duration, reaching over $0.60/image at 4k and up to $0.80/second for FLUX 3 Video
  • The vendor's pricing calculator has shown different default figures than its published 'from' prices for some FLUX.2 tiers
  • Open weights carry license restrictions: FLUX 3 Image Klein's free commercial use applies only to companies with annual revenue under $5M, FLUX.1 [dev] remains non-commercial, and larger-scale self-hosting needs a separate paid license
  • A growing lineup across FLUX 3, FLUX.2, and legacy FLUX.1 tiers can be confusing when choosing a model

Choose FLUX if…

  • You need photorealistic image or video generation through an API
  • You want consistent characters or styles across images using multiple references
  • You want a self-hostable model, such as FLUX 3 Image Klein, built to run on consumer GeForce RTX GPUs with as little as 8GB VRAM
  • You need text rendering inside generated images
  • You prefer per-image or per-second billing over a monthly subscription

Frequently Asked Questions

Is FLUX better than Stable Diffusion?
Black Forest Labs' Flux models, now including FLUX 3, are generally regarded as producing sharper, more photorealistic results with better text and hand rendering than SDXL. Stable Diffusion still has a larger community ecosystem of fine-tuned models, LoRAs, and tools built around it.
Can I run FLUX locally?
Yes. Open-weight variants such as FLUX.1 [schnell], FLUX.2 [dev], FLUX.2 [klein], and FLUX 3 Image Klein (available on Hugging Face as black-forest-labs/flux-3-image-klein and -klein-base) can be run locally via ComfyUI or Diffusers, including on consumer GeForce RTX GPUs with as little as 8GB VRAM; other VRAM requirements vary by model size and quantization.
What is FLUX.1?
FLUX.1 was Black Forest Labs' original model generation, released in August 2024, offered in Schnell (fast, open weights), Dev (open weights, non-commercial), and Pro (API only) tiers. It has since been superseded by FLUX.2 and FLUX 3, though FLUX.1 tiers such as Kontext and Pro Ultra remain listed and available.
Also consider
ChatGPT Images
Image generation and precise editing inside ChatGPT, text and UI rendering
ComfyUI
open-source AI image generation, node workflows, local GPU, 4K video
Ideogram
AI image generation, text-in-image, canvas editing
User Reviews

Leave a Review

★★★★★

Reviews are published after moderation. We don't share your email.

No reviews yet — be the first to share your experience.