US Startup BlackForestLabs Launches Text-to-Image AI Model FLUX.1 with Prominent Features
TapTechNews August 12th news, the US startup BlackForestLabs launched the AI text-to-image model FLUX.1 on August 1st. This model is said to be prominent in 'latent diffusion, stable diffusion, and adversarial diffusion distillation', and can instantly generate various images according to user prompts.
The official compared this model with other competitors' products, and the results showed that the Elo score (TapTechNews note: The Elo scoring system is a scoring method widely used in competitive games such as chess, mainly used to calculate the strength level of the opponent in the competition) of the series models 'surpassed' StableDiffusion3Ultra, Ideogram, Midjourney6.0, DALL・E3 and other competing products.
TapTechNews learned that FLUX.1 currently provides three versions to meet the needs of different users (project address click here to visit), users can try these models through different platforms to explore their performance in practical applications. The three models are as follows:
FLUX.1pro: This model is positioned as the commercial version, supporting enterprise-customized solutions, emphasizing high-precision prompt following, excellent visual quality, rich image details, and diverse output formats.
FLUX.1dev: This model is positioned as the scientific research (open source) version, mainly for non-commercial purposes, with open weights, suitable for research and developers to use.
FLUX.1schnell: This is a'small model', a streamlined version based on FLUX.1, suitable for local development and personal use, and adopts the Apache2.0 license.
The official said that all versions of the models support generating images in different proportions, and the highest supports generating images with 2 million pixels. Although currently FLUX.1 only supports generating images through prompts, the official mentioned that they plan to expand the model in the future to allow the model to generate video clips through prompts.