SDXL Turbo

SDXL Turbo (Stability AI)

Ultra-fast text-to-image generation optimized for speed over perfection.

• Turbo diffusion. A distilled version of SDXL designed to produce usable images in 1–4 steps, drastically reducing latency compared to standard diffusion models.

• Real-time friendly. Built for interactive use cases where speed matters more than absolute image quality or fine detail.

• Lower compute cost. Runs faster and cheaper than full SDXL, making it suitable for rapid previews, UI generation, and iterative workflows.

• Prompt-light behavior. Best results with simple prompts; less suited for long, highly detailed instructions.

• Open weights. Released under a permissive Stability AI license and widely supported in the Diffusers ecosystem.

Why pick it for Norman AI?

SDXL Turbo is ideal when instant feedback matters. It’s a strong choice for live generation, quick previews, and UX-driven image workflows where waiting on diffusion isn’t an option.

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