Wan2.2-T2V-A14B

Wan2.2-T2V-A14B (Wan-AI)

Large-scale text-to-video generation built for visual consistency and detailed motion.

• Text-first pipeline. Generates short videos directly from text prompts, focusing on scene coherence, subject stability, and structured motion over time.

• Heavy diffusion backbone. ~14B-parameter diffusion model that prioritizes output quality and temporal consistency over speed or low compute.

• Stable results. Better at maintaining characters, layouts, and motion across frames compared to smaller or adapted video models.

• Diffusers-ready. Ships in the Diffusers format, making it easier to integrate into existing video generation workflows.

• Research-oriented release. Designed for experimentation and evaluation, not lightweight or real-time use.

Why pick it for Norman AI?

Wan2.2-T2V-A14B is a strong option when you want to benchmark or explore high-end text-to-video quality. It’s well suited for comparing prompt behavior, motion realism, and scene continuity in large video models.

response = await norman.invoke(
    {
        "model_name": "wan2.2-t2v-a14b",
        "inputs": [
            {
                "display_title": "Prompt",
                "data": "A cat playing with a ball on mars"
            }
        ]
    }
)