The release of Flex.1 alpha by OpenFlux’s creator, ostris, marks a significant milestone in the world of AI-generated imagery. With an upgraded 8 billion parameters, this new version offers impressive advancements in image generation, thanks to its fine-tuning and improved data training.
What's more, Flex.1 alpha is now open-source under the Apache 2.0 license, making it available for both commercial and personal use. This blog explores the features of Flex.1 alpha and how it integrates seamlessly into Shakker AI, a powerful tool for working with Flux models.
How Shakker AI Integrates Flex.1 Alpha for Enhanced AI Workflows
The Journey of Flex.1 Alpha: From FLUX.1 to Flex.1
Why Use Flex.1 Alpha with Shakker AI?
Released recently, Flex.1 alpha is a pre-trained rectified flow transformer model with 8 billion parameters capable of generating stunning images from text prompts. Unlike its predecessor, the OpenFlux model, Flex.1 alpha benefits from more extensive data training, significantly enhancing its performance.
This new model is based on the FLUX.1-dev architecture, but with reduced complexity—specifically, the MMDiT blocks have been cut from 19 to just 8, reducing the total number of parameters. Originally fine-tuned from the FLUX.1-schnell, Flex.1 alpha retains the Apache 2.0 license, making it freely available for anyone to use.
Notably, Flex.1 alpha includes a unique guidance embedder that eliminates the need for CFG to generate high-quality images, further streamlining its capabilities for AI-driven artwork creation.
Key Features of Flex.1 Alpha:
For a detailed exploration of Flex.1 Alpha’s capabilities, visit Flex.1 Alpha on HuggingFace.
Shakker AI provides a robust platform that integrates Flux models, including Flex.1 alpha, with the Stable Diffusion A1111 WebUI and ComfyUI, among other tools. This integration enables creators to generate, edit, and fine-tune images effortlessly.
Shakker AI Canvas Generator offers powerful features like:
The flexibility of Shakker AI combined with Flex.1 alpha empowers users to create everything from realistic photography to stylized anime images. The seamless integration with ComfyUI and A1111 WebUI makes it a highly adaptable tool for a variety of creative workflows.
Flex.1 alpha began as a FLUX.1-schnell-training-adapter, designed to fine-tune LoRA models for use with FLUX.1-schnell. Over time, ostris refined this adapter into a standalone base model, OpenFLUX.1, which underwent months of additional training and pruning to create a more efficient version: Flex.1 alpha.
The model’s architecture has been optimized for flexibility, making it much more than just a simple text-to-image generator. It can be fine-tuned easily, with users being able to bypass the guidance embedder for greater control over the model’s output.
Integrating Flex.1 alpha into Shakker AI enables an array of enhanced features that allow users to:
The introduction of Flex.1 alpha opens new doors for AI-driven creativity, particularly when integrated into Shakker AI’s powerful platform. With its fine-tuning capabilities and advanced model training, Flex.1 alpha is set to revolutionize the way AI-generated images are created and refined. Whether you are working on art, photography, or game design, the integration of Flex.1 alpha with Shakker AI offers unparalleled flexibility and performance.
Explore Shakker AI today and unlock the potential of Flex.1 alpha for your creative projects.