FLUX.1: Next-Generation Image Generation
FLUX.1 is Black Forest Labs' image generation model, created by the original architects of Stable Diffusion. It delivers state-of-the-art image quality with superior prompt adherence and text rendering.
Specifications
At a glance
Parameters
12B (Pro/Dev) / ~4B (Schnell)
Architecture
Rectified Flow Transformer
Release Date
August 2024
Variants
Pro (closed) / Dev (open) / Schnell (open, fast)
Resolution
Up to 2048x2048
Licence
Apache 2.0 (Schnell) / Non-commercial (Dev) / Proprietary (Pro)
Overview
About FLUX.1
FLUX.1 is the debut model from Black Forest Labs, founded by the key researchers behind Stable Diffusion (including Robin Rombach). It represents a generational leap in image generation, using a rectified flow transformer architecture that delivers superior image quality, prompt adherence, and text rendering compared to previous diffusion models. FLUX.1 comes in three variants: Pro (the highest quality, available via API), Dev (open weights for non-commercial use, strong quality), and Schnell (open weights under Apache 2.0, optimised for speed). The Pro and Dev models use 12B parameters — significantly larger than Stable Diffusion's models — resulting in more detailed, coherent, and aesthetically pleasing generations. The model excels at rendering text within images, complex compositional scenes, and photorealistic content. It has quickly become one of the most popular image generation models, with broad integration across platforms like Replicate, Together AI, and ComfyUI. FLUX.1 represents the natural evolution of the open image generation ecosystem that Stable Diffusion pioneered.
Strengths
Capabilities
- State-of-the-art image quality with strong aesthetic appeal
- Superior text rendering within generated images
- Excellent prompt adherence for complex scene descriptions
- Three variants balancing quality, speed, and licensing
- Schnell variant under Apache 2.0 for commercial use
- ComfyUI and community workflow integration
- LoRA fine-tuning support for custom styles
Considerations
Limitations
- Pro variant is API-only with per-image costs
- Dev variant restricted to non-commercial use
- Larger model size (12B) requires more VRAM than Stable Diffusion
- Newer ecosystem with fewer extensions than Stable Diffusion
- No native video generation capabilities
Best For
Ideal use cases
- Commercial image generation requiring high quality and text rendering
- Marketing and advertising visual content creation
- Product mockup and concept visualisation
- Creative workflows requiring strong prompt adherence
- Migration from Stable Diffusion for improved quality
Pricing
Schnell: Free (Apache 2.0, self-hosted). Dev: Free for non-commercial use. Pro: API pricing varies by provider (~$0.03–$0.06 per image). Available via Black Forest Labs API, Replicate, Together AI, and others.
FAQ
Frequently asked questions
FLUX.1 is created by the same core team that built Stable Diffusion. It delivers significantly better image quality, prompt adherence, and text rendering. However, Stable Diffusion has a much larger ecosystem of extensions, fine-tuned models, and community tools.
For commercial use: Schnell (free, Apache 2.0) for speed or Pro (API, paid) for maximum quality. For personal/research: Dev offers the best quality with open weights. Most commercial products start with Schnell and upgrade to Pro where quality matters most.
Yes. FLUX.1 is one of the best models for rendering readable text within images, making it effective for mockups, social media graphics, and any use case requiring text as part of the generated image.
The 12B Dev/Pro models require 12GB+ VRAM for comfortable generation (NVIDIA RTX 4070 or better). Schnell is more efficient and can run on 8GB+ VRAM GPUs. Quantised versions reduce requirements further.
Need help with FLUX.1?
Our team can help you evaluate and implement the right AI tools. Book a free strategy call.