FLUX.2-dev

FLUX.2-dev (Black Forest Labs)

Next-generation text-to-image generation focused on prompt accuracy and visual consistency.

• Evolved foundation model. Successor to the FLUX.1 series, trained to improve prompt understanding, composition, and semantic alignment across complex scenes.

• Cleaner outputs. Produces sharper images with better anatomy, lighting, and fewer random artifacts compared to earlier FLUX and SDXL-class models.

• Prompt-faithful by design. Handles long, structured prompts with more predictable results, making iteration and comparison easier.

• Heavy but capable. Requires more VRAM and compute than classic diffusion models, trading efficiency for higher visual quality.

• Dev release. Published as a development checkpoint for research and evaluation, not a locked production model.

Why pick it for Norman AI?

FLUX.2-dev is a strong option when you want to explore the next step beyond SDXL and FLUX.1. It’s well suited for testing prompt behavior, comparing image quality, and evaluating future text-to-image workflows.

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