FLUX.2 [klein] 4B

Image modelOpen weightsCard summarized

The fastest, most lightweight FLUX.2 model, with open weights that run on consumer GPUs.

Source: FLUX.2 [klein] 4B model card

Use it via API

$0.014 / image

Code example Compare

Where to call it

BFL APIflux-2-klein-4b
Self-host (weights)black-forest-labs/FLUX.2-klein-4B

Supports

Image input

Full feature support

Good for

Sub-second image generation
Interactive workflows, production and latency-critical apps
Text-to-image and multi-reference editing in one model

Overview

Black Forest Labs · 03klein

klein 4B

The fastest, most lightweight FLUX.2 model, with open weights that run on consumer GPUs.

Typical price

$0.014 / image

Text-to-image, 1 MP (e.g. 1024×1024)

Max resolution: 4 MP (FLUX.2 family limit)
Aspect ratios: —
Modes: text-to-image, image editing
Released: 15 Jan 2026

FLUX.2 [klein] 4B is a 4 billion parameter rectified flow transformer for text-to-image generation and multi-reference editing, released under Apache 2.0. It is step-distilled to 4 steps for sub-second inference and fits in about 13 GB of VRAM. Model card intro

flux-2-klein-4b

Typical price: $0.014 / image
Max resolution: 4 MP (FLUX.2 family limit)
Weights: Open · Apache 2.0
Released: 15 Jan 2026

text-to-image
image editing
multi-reference editing (up to 4)
LoRA inference (beta)