FLUX.2 [klein] Base 4B is a compact, open-weight image generation model developed by Black Forest Labs. Released as part of the FLUX.2 family, the "[klein]" variant is designed for high-speed, interactive visual intelligence. The model utilizes a rectified flow transformer architecture that unifies text-to-image synthesis and image editing into a single checkpoint. As an undistilled base model, it preserves the full training signal, offering higher output diversity and greater flexibility for fine-tuning and LoRA training compared to its distilled counterparts.
Technically, the model integrates vision-language understanding—reportedly utilizing a Qwen3B backbone for prompt processing—to improve spatial relationships, material properties, and compositional logic. It supports high-fidelity outputs at resolutions up to 4 megapixels and introduces advanced control features, including JSON Structured Prompts for granular specification of scene elements and support for specific HEX codes for precise color reproduction.
The [klein] 4B Base model is capable of multi-reference composition, allowing users to combine multiple input images to guide a novel output. Despite its compact 4-billion parameter size, it maintains frontier performance in photorealism and typography. The model is released under an Apache 2.0 license, making it suitable for both commercial and research applications. It is optimized for local deployment on consumer-grade hardware, requiring approximately 13GB of VRAM to operate.