Released yesterday, the klein series includes two primary parameter counts: 4 billion (4B) and 9 billion (9B). The model weights are available on Hugging Face and the code is on Github, according to BFL.

While the larger models in the FLUX.2 family (max and pro), released in November 2025, aim for photorealism and "grounding search" capabilities, klein is specifically designed for consumer hardware and latency-critical workflows. This makes it suitable for applications where quick image generation is essential, such as real-time content creation and interactive design tools.

The 4B version is available under an Apache 2.0 license, allowing organizations and developers to use the klein models for commercial purposes without licensing fees to BFL or intermediaries. This open-source approach could democratize access to AI image generation technology, enabling wider adoption across various industries. Several AI image and media creation platforms, including Fal.ai, already support the new models.

BFL, founded by former Stability AI engineers, is building a suite of open-source AI image generators. The release of FLUX.2 klein represents a continued effort to provide accessible and efficient AI tools for image creation. The company hopes that the open-source nature of the 4B model will foster innovation and collaboration within the AI community.

The development of smaller, faster AI models like FLUX.2 klein addresses a key challenge in the field: the high computational cost associated with running large AI models. By optimizing for speed and lower compute requirements, BFL aims to make AI image generation more accessible to a wider range of users and devices. The implications of this development could be significant, potentially enabling AI-powered image creation on smartphones, laptops, and other consumer devices.