Unified generation and editing
According to the model card, FLUX.2 klein unifies generation and editing in a compact architecture, with the 9B variant supporting text-described generation and multi-reference editing.
Open Source Model Profile · black-forest-labs
FLUX.2-klein-9B is a 9.08B-parameter image model from black-forest-labs. According to the model card, it unifies text-described generation and editing in a distilled 4-step design.
FLUX.2-klein-9B is published by black-forest-labs as an image-to-image model. Safetensors metadata reports 9,078,581,248 parameters, about 9.08B. According to the model card, it is a 9-billion-parameter rectified flow transformer for generation from text with multi-reference editing.
According to the model card, FLUX.2 klein unifies generation and editing in a compact architecture, with the 9B variant supporting text-described generation and multi-reference editing.
According to the model card, the model combines a 9B flow model with an 8B Qwen3 text embedder, step-distilled to 4 inference steps at 1024x1024 and guidance 1.0.
According to the model card, the model is available via the BFL API and in ComfyUI and Diffusers through Flux2KleinPipeline.
According to the model card, the publisher describes pre-training filtering, targeted fine-tuning, inference filters, and pixel-layer watermarking with C2PA metadata.
Source: black-forest-labs/FLUX.2-klein-9B
Captured: Unknown. Processed: 2026-09-07T19:34:41.503226+00:00.
The FLUX.2 [klein] model family are our fastest image models to date. FLUX.2 [klein] unifies generation and editing in a single compact architecture, delivering state-of-the-art quality with end-to-end inference in as low as under a second . Built for applications that require real-time image generation without sacrificing quality. FLUX.2 [klein] 9B is a 9 billion parameter rectified flow transformer capable of generating images from text descriptions and supports multi-reference editing capabilities. Our flagship small model. Defines the Pareto frontier for quality vs. latency across text-to-image, single-reference editing, and mul…
F001F002F003F004F005F006F008F009F010F011F013F014F015F016F017F018F020F024F025F026F027