Seedance 2.5 Is Now on VIDEO AI ME: What's New

Seedance 2.5 is live in the VIDEO AI ME editor. It sits in the same model picker as Sora 2, Kling, Seedance 2.0 Fast, Happy Horse, Grok Imagine 1.5 and Gemini Omni Flash, and you reach it in a browser, in English, on a normal subscription. This post covers what actually shipped, what it changes for ad creative, and the cases where staying on Seedance 2.0 is still the correct call.

Some context on why this matters. ByteDance released Seedance 2.5 on 2026-07-31 inside Jimeng AI and Doubao, which in practice expect a Chinese phone number and a Chinese language interface, and announced the model on the ByteDance Seed site. API access was described as arriving later through BytePlus ModelArk. So most of the people searching for this model right now cannot reach it at all. That is the gap this release closes.

That clip is the same 15 second spot run through both models, 2.5 in the top panel and 2.0 below, from an identical prompt. The spot opens on a woman walking on the ocean at sunset, cuts to a founder at a laptop, and ends on the brand card. Watch it with sound on, because the audio is part of what changed.

What shipped with Seedance 2.5

Three generation modes, selected automatically by how many reference images you attach:

The rest of the specification, in one list:

• Durations: 4, 8, 12, 15, 20, 25 and 30 seconds. A 30 second clip is generated in a single pass, not stitched from shorter renders.
• Resolutions: 480p and 720p.
• Aspect ratios: auto, 21:9, 16:9, 4:3, 1:1, 3:4 and 9:16. Image to video always follows the input image, so the aspect picker does not apply in that mode.
• Audio: always generated, always included in the price. Sound effects, ambient sound and lip synced speech. No toggle, no surcharge.

One point of precision, because the launch coverage blurs it. The Seedance 2.5 model also accepts video references and audio references, a much larger reference image count, and editing controls that operate on specific timestamps. On VIDEO AI ME today you get text to video, image to video, and reference to video with up to four images. Plan your work around that list, not around the model announcement.

What Seedance 2.5 changes for ad creative

If you produce paid social creative, three things change in a way you can feel in a production week.

A 30 second spot is now one render. Previously a half minute of coherent video meant generating four to six short clips, hoping the subject looked the same in each, and assembling them. A single pass 30 second generation removes the assembly and removes the continuity gamble. The lighting does not jump, the wardrobe does not change, and the room stays the same room.

Audio arrives with the picture. Dialogue, ambient sound and effects are produced in the same render, and the speech is lip synced. That deletes a whole stage of the pipeline for testimonial style and spokesperson style ads: no separate text to speech step, no manual sync pass, no scramble to find a room tone bed that matches the shot.

Reference to video protects the two things buyers care about. Attach the creator in @Image1 and the product in @Image2 and bind them in the prompt by name. That is how you keep a face consistent across a batch of variants and keep the label on the bottle legible, which is the difference between a clip you can run and a clip you throw away.

The practical consequence is fewer steps per asset. Fewer steps means more variants per day at the same headcount, which is the only lever that reliably improves creative performance on Meta and TikTok, where refresh rate is a real constraint.

Who should switch from Seedance 2.0, and who should not

Being blunt about this saves you credits.

Switch to 2.5 when:

• Your ad needs spoken dialogue, especially a talking spokesperson or a testimonial.
• The concept runs longer than about eight seconds and has more than one beat.
• The same person or the same product must appear consistently across several shots or several variants.
• The idea depends on audio timing, like a line landing on a cut.

Stay on Seedance 2.0 Fast when:

• You are testing twenty hooks in a morning and only need to see whether an idea reads at all.
• The clip is silent b roll, a loop, or a background plate under a text overlay.
• You already have a prompt library that produces reliable output and no reason to rebuild it. If you are new to the family, our explainer on what Seedance 2.0 is sets the baseline, and the side by side of Seedance 2.5 against Seedance 2.0 shows where each one wins.

Most teams end up running both. The cheap model explores, the expensive model finishes.

What Seedance 2.5 costs on each plan

Credits are the billing unit and 1 credit equals 1 cent of generation cost. Seedance 2.5 is 23 credits per second at 480p and 48 credits per second at 720p. Reference to video costs the same per second as the other two modes.

Against the plans:

Read the arithmetic honestly. A full 30 second clip at 720p costs 1,440 credits, which is more than the entire Starter monthly allowance of 1,400. Thirty second work at 720p starts at Pro. Starter handles 30 second clips at 480p, at 690 credits each, comfortably. Pro buys roughly 116 seconds of 720p or 243 seconds of 480p per month, and Premium roughly 250 seconds of 720p or 521 seconds of 480p.

The habit that makes those numbers go further: explore at 480p, finish at 720p. Three 8 second exploration passes cost 552 credits at 480p. The same three at 720p cost 1,152. You are paying for information in the first case and for delivery in the second, and only the second needs the resolution. The full Seedance 2.5 pricing breakdown works through more of these cases, and current plan details live on the pricing page. There is no free tier for video generation, so an active subscription is required.

Try it on one real asset first

Pick an ad you already run and rebuild it, rather than starting with a blank prompt. You know what the finished thing should look like, so you can judge the output in seconds instead of arguing with yourself about taste.

A reasonable first run: 8 seconds, 480p, 9:16, one spoken line in quotes, a named camera move, and a named grade. Something like the opening shot of the demo spot above works as a template.

Aerial drone orbit, 45 degree down angle, slowly descending, counter clockwise. A woman in a flowing crimson silk gown walks barefoot on the surface of a mirror calm ocean at golden hour, toward a low molten sun. Slow concentric ripples radiate from each step. Slow motion, anamorphic lens flare, teal and orange cinematic grade, rich blacks, gentle halation. Epic luxury commercial style, 16:9, photoreal.

Then read the result with the sound on, change exactly one variable, and run it again. The step by step guide to using Seedance 2.5 covers the full loop including reference images and the settings that matter most.

Frequently Asked Questions

Is Seedance 2.5 available on VIDEO AI ME right now?

Yes. Seedance 2.5 appears in the model picker in the /new editor alongside Sora 2, Kling, Seedance 2.0 Fast, Happy Horse, Grok Imagine 1.5 and Gemini Omni Flash. It runs in the browser, in English, on any active subscription. There is no waitlist and no separate app to install.

Do I need a Chinese phone number or the Jimeng app to use Seedance 2.5?

Not on VIDEO AI ME. ByteDance shipped Seedance 2.5 inside Jimeng AI and Doubao, which in practice expect a Chinese phone number and a Chinese language interface. On videoai.me you sign in with a normal account and pick the model from a dropdown.

How much does one Seedance 2.5 clip cost?

Generation is billed in credits, where 1 credit equals 1 cent of generation cost. Seedance 2.5 costs 23 credits per second at 480p and 48 credits per second at 720p. An 8 second clip is 184 credits at 480p or 384 credits at 720p. A 15 second clip is 345 credits at 480p or 720 credits at 720p.

Does Seedance 2.5 generate audio, or do I still add a voiceover?

Audio is generated in the same pass as the picture and is always included in the price. That covers sound effects, ambient sound and lip synced speech. There is no audio toggle and no audio surcharge, so a finished clip arrives with its soundtrack already attached.

What is the longest Seedance 2.5 clip I can generate?

Thirty seconds. The duration menu offers 4, 8, 12, 15, 20, 25 and 30 seconds, and a 30 second clip is generated in a single pass rather than stitched together from shorter renders. That means one continuous scene with continuous audio.

Should I move all of my ad creative to Seedance 2.5?

No. Move the work that needs spoken dialogue, longer running time or subject consistency across a multi beat spot. Keep cheap silent b roll and high volume hook testing where it already works. Deciding per asset costs less than switching everything at once.