Wan 2.5

Wan 2.5 is Alibaba's video model from the Tongyi Wanxiang family, unveiled on September 24, 2025 as a preview release. It uses a native multimodal architecture across text, image, video, and audio, and it was the first model in the Wan series to generate synchronized human voice, sound effects, and visuals together in a single pass. On HeyOz it's wired straight into your brand kit and ad formats.

Pick Wan 2.5 as your model, write a prompt describing the scene and any dialogue you want spoken, and add a reference image if you're starting from a still. Generate, watch it back with sound, adjust, and run the next variant.

That's what HeyOz is built for. Point it at your product photo or write the scene, and the spot comes back on-brand, with sound, and sized for Meta, TikTok, and more.

You can start creating on HeyOz for free. See our pricing page for what's included on each plan.

Wan 2.5 generates clips of roughly 5 to 10 seconds, with the synchronized audio carried through the whole clip.

A text prompt, and optionally a reference image to animate for image-to-video. Because the model handles audio natively, you can write the dialogue into the prompt and have it spoken on screen — you don't need to supply a separate voice track.

Up to 1080p at 24fps, with lighter tiers available for drafting. Generate at 1080p for the cut you intend to run, and use a lower tier while you're testing angles.

Wan 2.5 introduced native synchronized audio to the family and generates from text or a single image. Wan 2.6, released later in December 2025, adds reference-to-video — casting a real person from a short clip of them — and longer generations up to 15 seconds. Pick 2.5 when you want fast text or image-to-video with sound; pick 2.6 when you're casting a specific spokesperson. Both are on HeyOz.