black-forest-labs/flux-1.1-pro
Faster, better FLUX Pro. Text-to-image model with excellent image quality, prompt adherence, and output diversity.
Pricing
black-forest-labs/
flux-1.1-pro
Pricing for Synexa AI models works differently from other providers. Instead of being billed by time, you are billed by input and output, making pricing more predictable.
For example, generating 100 images should cost around $2.00.
Check out our docs for more information about how per-request pricing works on Synexa.
| Provider | Price ($) | Saving (%) |
|---|---|---|
| Synexa | $0.0200 | - |
| replicate | $0.0400 | 50.0% |
| fal | $0.0500 | 60.0% |
Readme
FLUX1.1 pro marks a significant advancement in image generation technology, delivering a remarkable six times faster generation speed compared to its predecessor, FLUX.1 pro. This new iteration not only accelerates workflows but also elevates the quality of generated images, demonstrating improved prompt adherence and greater diversity in outputs.
Key Features and Enhancements
FLUX1.1 pro offers a suite of improvements designed to optimize the image generation experience:
- Unprecedented Speed and Efficiency: Experience significantly faster generation times and reduced latency. FLUX1.1 pro is engineered to provide an optimal balance between exceptional image quality and rapid inference speed, streamlining creative processes and boosting productivity.
- Elevated Performance and Image Quality: Under the development codename “blueberry,” FLUX1.1 pro has undergone rigorous testing and validation within the Artificial Analysis image arena, a leading platform for benchmarking text-to-image models. Demonstrating its superior capabilities, FLUX1.1 pro has surpassed all competing models on the leaderboard, achieving the highest overall Elo score. This achievement underscores its exceptional performance and positions it at the forefront of text-to-image generation technology.
For a comprehensive overview of this release, read the official announcement from Black Forest Labs here.
Technical Architecture
The foundation of all public FLUX.1 models is built upon a sophisticated hybrid architecture combining:
- Multimodal Architectures: multimodal for robust feature extraction and representation.
- Parallel Diffusion Transformer Blocks: Parallel diffusion transformer to enable efficient and high-quality image generation.
These architectural components are scaled to 12 billion parameters, allowing for complex and nuanced image synthesis. FLUX.1 models achieve advancements over previous state-of-the-art diffusion models through the implementation of:
- Flow Matching: Leveraging flow matching, a conceptually clear and versatile method for training generative models, with diffusion as a specialized instance.
- Rotary Positional Embeddings: Incorporating rotary positional embeddings to enhance model understanding of spatial relationships and improve performance.
- Parallel Attention Layers: Utilizing parallel attention layers to boost model performance and optimize hardware efficiency.
License Information
By utilizing FLUX.1 pro via Replicate, you are agreeing to the terms outlined in the Black Forest Labs API agreement and the Black Forest Labs Terms of Service. Please review these documents carefully before use.