FLUX.2 [klein] 9B

FLUX.2 [klein] 9B (Black Forest Labs)

The "fast" version of FLUX.2, built for real-time speed without sacrificing professional-grade quality.

• Speed-tuned architecture. A 9-billion parameter model distilled to run in just 4 steps, delivering sub-second image generation on high-end GPUs.

• Unified workflow. Handles text-to-image, image-to-image, and multi-image blending (up to 5 images) within a single model—no need for extra plugins or separate checkpoints.

• High prompt fidelity. Inherits the FLUX.2 brain, meaning it understands complex descriptions and renders sharp, accurate text better than standard SDXL models.

• Production efficiency. Optimized to sit on the "sweet spot" of the performance curve, offering high-fidelity results with significantly lower latency than the "Dev" or "Pro" versions.

• Hardware friendly. While it is a heavy transformer, its distilled nature allows for rapid iteration on consumer-grade cards like the RTX 4090.

Why pick it for Norman AI?

FLUX.2 [klein] 9B is the go-to choice when latency matters. It’s the best option for building interactive tools, real-time creative assistants, or any application where you need high-quality visuals to appear almost instantly. Use it when you want the power of FLUX.2 without the long wait times of larger models.

response = await norman.invoke(
    {
        "model_name": "flux.2-dev",
        "inputs": [
            {
                "display_title": "Prompt",
                "data": "A cat playing with a ball on mars"
            }
        ]
    }
)

response = await norman.invoke(
    {
        "model_name": "flux.2-dev",
        "inputs": [
            {
                "display_title": "Prompt",
                "data": "A cat playing with a ball on mars"
            }
        ]
    }
)

response = await norman.invoke(
    {
        "model_name": "flux.2-dev",
        "inputs": [
            {
                "display_title": "Prompt",
                "data": "A cat playing with a ball on mars"
            }
        ]
    }
)