AuraFlow, the latest open-source AI model in the Aura series, has been launched by @cloneofsimo and @FAL. AuraFlow v0.1 is a flow-based text-to-image generation model that boasts 6.8 billion parameters, making it the largest open-sourced model of its kind. The model features DiT Encoder blocks and improved instruction following, achieving a GenEval score of 0.703 with prompt enhancement. AuraFlow supports 1024x1024 resolution and is available under the Apache-2.0 license. It can be accessed via Pixel Dojo. The release has been well-received, with users highlighting its capabilities in generating high-quality images and its potential in various creative applications. Congratulations to Simo, Batuhan, and the Fal AI team on this great release.