FLUX.2 [klein] 4B

Compact 4-billion parameter text-to-image model from Black Forest Labs. Step-distilled to 4 inference steps for sub-second generation. Unifies text-to-image and image editing in a single architecture with Apache 2.0 licensing.

By Black Forest Labs

Pricing: $0.005 per megapixel

Overview

FLUX.2 [klein] 4B is a compact, Apache 2.0-licensed image generation model from Black Forest Labs. At 4 billion parameters and just 4 inference steps, it delivers sub-second generation without sacrificing quality. It uses a Mistral 3 VLM text embedder and supports up to 4 megapixel output resolution.

Key Features

• **Sub-second inference** — distilled to 4 steps for real-time applications
• **Multi-reference editing** — accepts up to 4 input images for guided generation
• **Text rendering** — strong text-in-image capabilities inherited from FLUX.2
• **Transparent backgrounds** — native RGBA PNG output support
• **Apache 2.0** — fully commercial use permitted

Architecture

Rectified flow transformer with FLUX.2 VAE for pixel-to-latent conversion. The distilled variant uses `guidance_scale=1.0` by default. Requires ~13 GB VRAM (fits on RTX 3090/4070+).

When to Use

Ideal for high-volume production workloads, real-time applications, consumer-facing products, and edge deployment where speed and commercial licensing matter.

Examples

• A vintage travel poster for 'RUNFLOW AIRWAYS' featuring a retro DC-3 airplane flying over snow-capped mountains at sunset, bold art deco typography, warm orange and teal color palette, textured paper grain, 1940s airline advertising style

Frequently asked questions

What is FLUX.2 Klein 4B?

FLUX.2 [klein] 4B is a compact 4-billion parameter image generation model from Black Forest Labs. It's distilled to just 4 inference steps for sub-second generation while maintaining high image quality. It uses a Mistral 3 VLM text embedder and supports up to 4 megapixel resolution.

What license does Klein 4B use?

FLUX.2 [klein] 4B uses the Apache 2.0 license, making it fully available for commercial use with no restrictions. This is notable because the larger 9B variant uses a non-commercial license.

How fast is Klein 4B compared to other FLUX models?

Klein 4B generates images in sub-second time thanks to step distillation (4 steps vs 25-50 for standard models). It's significantly faster than FLUX.2 Pro or FLUX.2 Max while trading some quality for speed — similar to FLUX.2 Flash but with a smaller footprint.

Can Klein 4B do image editing?

Yes. Klein 4B unifies text-to-image generation and image editing in a single architecture. You can pass up to 4 reference images alongside your text prompt for guided generation, style transfer, or editing tasks. Use the edit endpoint for editing workflows.

What resolution does Klein 4B support?

Klein 4B supports output up to 4 megapixels (e.g., 2048x2048). The recommended resolution is around 1-2 megapixels for optimal quality. All dimensions must be multiples of 16.

What's the difference between Klein 4B and Klein 9B?

The 4B variant is smaller (13GB VRAM), Apache 2.0 licensed, and uses a Mistral 3 VLM text embedder. The 9B variant is larger (29GB VRAM), non-commercial licensed, uses an 8B Qwen3 text embedder, and produces slightly higher quality outputs with more diversity.

How much does FLUX.2 Klein 4B cost on Runflow?

FLUX.2 Klein 4B costs $0.014 per megapixel for the first megapixel and $0.001 for each additional megapixel. A standard 1024x1024 image (~1MP) costs approximately $0.014.

Related models

• FLUX 1.1 [pro] Ultra, Ultra-resolution variant of FLUX 1.1 [pro] with 4MP+ outputs and richer detail. Drop-in upgrade when the standard FLUX 1.1 [pro] resolution is the bottleneck.
• FLUX 1.1 [pro], Black Forest Labs' production-grade FLUX 1.1. State-of-the-art prompt adherence and photoreal quality at competitive per-image pricing.
• SDXL-Lightning, ByteDance lightning-fast SDXL distillation. 4-step text-to-image at near-real-time latency.
• GPT Image 2, GPT Image 2.0, OpenAI's latest image model, is capable of creating extremely detailed images with fine typography.

Discoverable surfaces

• Dispatch endpoint: POST https://api.runflow.io/v1/models/flux-2-klein-4b/runs
• Per-model spec (markdown): https://app.runflow.io/models/flux-2-klein-4b/llms.txt
• Docs page: https://docs.runflow.io/models/flux-2-klein-4b
• Public OpenAPI spec: https://docs.runflow.io/api/openapi.public.json
• Agent skill (start here): https://www.runflow.io/.well-known/agent-skills/runflow/SKILL.md