FLUX.1 Dev

FLUX.1 Dev is a 12 billion parameter rectified flow transformer capable of generating high-quality images from text descriptions. It's designed for developers and researchers who want to explore and build upon advanced AI image generation technology.

FLUX.1 Dev stands out with its transformer-based flow architecture, offering superior prompt adherence, visual quality, and output diversity. It excels at generating detailed, photorealistic images while maintaining excellent control over composition and style.

FLUX.1 Dev is available under a non-commercial license. Generated outputs can be used for personal, scientific, and research purposes. For commercial applications, please check the licensing terms or consider FLUX.1 Pro.

Flux Dev is trained using guidance distillation technology, offering output quality second only to Flux Pro while being optimized for efficiency. It has open weights and is designed for non-commercial use with faster inference times. Flux Pro is the state-of-the-art model providing the highest quality output and is available only via API for commercial applications.

Flux Dev's image-to-image capability allows you to transform existing images using text prompts while preserving the core composition. You can control the transformation strength from 0.1 (subtle changes) to 1.0 (dramatic changes). It's perfect for style transfers, artistic conversions, and creative reimagining of existing images.

Recommended settings include: Guidance Scale of 3.5 for balanced creativity and prompt adherence, 28-50 inference steps for detailed images, and using seed values for reproducible results. For image-to-image, start with strength values of 0.7-0.8 for most transformations.

Flux Dev has superior text rendering capabilities compared to many other models. For best results, clearly specify the text content and its placement in your prompt, and consider using quotation marks for exact text. The model excels at creating readable text within images.

Flux Dev supports various resolutions up to 2.0 megapixels, including common aspect ratios like 1:1 (1024×1024), 16:9 (1344×768), 9:16 (768×1344), and custom dimensions. This flexibility makes it suitable for diverse creative applications.

For better image-to-image results: 1) Start with high-quality, clear source images, 2) Be specific about the target style in your prompts, 3) Experiment with strength values (0.1-0.3 for subtle changes, 0.7-0.8 for strong stylization), 4) Layer multiple style elements for complex transformations, and 5) Consider lighting and atmosphere changes in your prompts.