Black Forest Labs, the original team behind Stable Diffusion, announced the release of Flux, the largest state-of-the-art open-source text-to-image model available to date. Flux pushes the boundaries of creativity and performance with an impressive 12 billion parameters, offering aesthetics comparable to Midjourney.
Thanks to the integration of fal’s cutting-edge inference engine, you can run Flux models up to 2x faster than Eager Torch. This results in faster processing times while maintaining exceptional quality and detail.
Differences between Flux Variations: Dev, Schnell and Pro
Flux.1 [dev]
- License: Non-commercial, open-source.
- Performance: Based on a model distilled from the Flux Pro, it offers similar output quality but with greater efficiency. It is ideal for research and development purposes.
- Characteristics:
- Supports generating high-quality images.
- Designed for greater adherence to prompts, comparable to closed models.
- Accessible on Hugging Face, with weights available for use.
Flux.1 [schnell]
- License: Open-source, available under the Apache 2.0 license.
- Performance: Optimized for speed, it is designed for local use and personal projects. It can run up to 10 times faster than other standard models.
- Characteristics:
- Excellent for applications requiring rapid image generation.
- It maintains good visual quality, although it may be less detailed than the Pro model.
- Ideal for developers and hobbyists who want fast and flexible access.
Flux.1 [pro]
- License: Closed-source, available only via API.
- Performance: Considered the flagship model, it offers unmatched performance in terms of visual quality, detail and output diversity. It is designed for professional applications where quality is paramount.
- Characteristics:
- Maximum image generation quality, surpassing other models such as Midjourney and DALL·E 3.
- Excellent adherence to prompts and ability to generate highly detailed images.
- Access via API, making it suitable for commercial and professional integrations.
