ByteDance's Dreamina platform — and TikTok's ad tools — now run Seedance 2.5, with 30-second clips and up to 50 references.
Last verified · 2026-10-23 · by Moe Ameen
Published 2026-08-04 · Updated
Short answer
TikTok Dreamina Seedance 2.5 is ByteDance's Seedance 2.5 video model, which went live on Dreamina on July 31, 2026 and reached TikTok's Symphony ad tools on August 3. In Symphony, advertisers can generate 30-second clips, up from 15 seconds, and upload up to 50 image, video, and audio references, up from 9. The TikTok rollout covers select paid advertisers in select markets.
Key takeaways
Dreamina is ByteDance's AI creation platform for images and video, at dreamina.capcut.com. It runs multiple image and video models behind one editor, and it was the first place ByteDance's newest video model, Seedance 2.5, went live: the company announced Seedance 2.5's global launch debuting on Dreamina on July 31, 2026. On August 3, 2026, TikTok began rolling the same model into its advertiser toolkit through Symphony, so "TikTok Dreamina Seedance 2.5" is really one model reaching creators two ways — the Dreamina app and TikTok's ad creation flow.
The upgrade that matters for creators is length and control. In TikTok's Symphony integration, advertisers can now generate 30-second clips, up from 15 seconds previously, and upload up to 50 multi-modal references — images, video, and audio — up from 9, which gives far tighter control over look, motion, and consistency.
On Dreamina, the same model adds targeted editing, so you can revise selected parts of a clip instead of starting over. TikTok describes sharper, more realistic imagery with stronger preservation of lighting, motion, and character consistency across a shot.
Two things to keep straight. On the TikTok ad side, Seedance 2.5 is rolling out to select paid advertisers in select markets, so access is staged. And on Dreamina, model access and prices vary by subscription plan and region, so check current terms before you plan around it. Either way, the tool stops at the clip: it offers basic edits, but no branded captions for sound-off feeds, no persona system to keep one influencer identity consistent across many posts, and no scheduling or multi-platform publishing.
Dreamina can generate a virtual creator on camera, and that is exactly where its ceiling shows. A Dreamina render is a one-off: generate the same virtual creator twice and the face, wardrobe, and voice can drift, and the clip lands with no captions and no way to post it. If your plan is a recurring AI influencer — the same identity showing up in your feed every week — Kompozy is built for that gap.
Its AI Influencer persona pool holds one primary identity, and Gemini face-lock keeps that persona's face identical across every image and post, while HeyGen-powered Persona Shorts and Persona HeyGen generate the talking-head episodes Dreamina's scene generator does not. Use Dreamina for the eye-catching generated scene or product moment, then let Kompozy lock the persona, burn in on-style captions, reframe to 9:16, 1:1, or 16:9 per platform, and publish the same short across TikTok, Reels, YouTube Shorts, Facebook, LinkedIn, X, Pinterest, and Threads from one queue.
The second win is turning one Dreamina clip into a week of posts. Feed the concept into Kompozy and it fans it into a carousel breaking down the pitch, a face-locked Persona Photo, a quote card from the script, a blog draft, and platform-native captions in your voice through the Persona Brief. Dreamina makes the render; Kompozy makes it a consistent, scheduled, multi-format campaign.
A Persona short rendered in Kompozy with a recurring avatar. It shows the format only and was not generated with Seedance.
Best for:
Not ideal for:
It is ByteDance's Seedance 2.5 video model running in two places: Dreamina, ByteDance's creation platform at dreamina.capcut.com, which launched it on July 31, 2026, and TikTok's Symphony advertiser tools, which began rolling it out on August 3, 2026. It generates 30-second clips and accepts up to 50 multimodal references.
Advertisers can now generate 30-second clips, up from 15 seconds previously, and upload up to 50 multi-modal references (image, video, audio), up from 9. TikTok describes sharper imagery and stronger preservation of lighting, motion, and character consistency.
Creators use it on Dreamina, where model access and prices vary by subscription plan and region, so confirm current limits and pricing on the Dreamina site. On the TikTok ad side, Seedance 2.5 is rolling out to select paid advertisers in select markets.
It can generate a virtual-creator clip, but each render is a one-off and the identity can drift between generations. To keep one AI influencer's face and voice consistent across many posts, use Kompozy's persona pool and Gemini face-lock, then publish the clips across platforms.
Dreamina generates the clip but does not publish it. Bring the export into Kompozy to add branded captions, reframe per platform, lock your persona, fan it into a carousel and captions in your voice, and schedule and publish across TikTok, Reels, YouTube Shorts, X, LinkedIn, and more from one queue.
We checked every figure and vendor claim on this page against the sources listed below on 2026-10-23, including clip length, reference counts and access limits in Social Media Today's report and the July 31 launch in ByteDance's PR Newswire release; we removed 4K, credit and Jimeng claims we could not confirm. Kompozy plan and product details come from kompozy.io and kompozy.io/pricing.
Sources