What Is Seedance 2.5? Everything We Know So Far (Coming Soon to WaveSpeedAI)

Seedance 2.5 is ByteDance’s next-generation video model, and it is one of the most anticipated releases of the year. It has been announced and widely covered by media, but ByteDance has not published full specs yet — so everything below is early reporting, not final documentation.

Here is the short version: Seedance 2.5 is coming soon to WaveSpeedAI. You can bookmark the Seedance 2.5 API page now, and start building today on the live Seedance 2.0 API — when 2.5 ships, it will run through the same API and integration you already have.

TL;DR

• Seedance 2.5 has been announced as ByteDance’s follow-up to Seedance 2.0 and widely reported by media channels.
• The two headline upgrades: native 30-second clips in a single pass and up to 50 multimodal reference inputs per request.
• Reports say those 30-second clips can render in 4K.
• ByteDance has not released official specs, so treat the numbers here as early reporting until the model ships.
• Seedance 2.5 will be available on WaveSpeedAI — same unified API, pay-as-you-go pricing, no subscription. Track it on the Seedance 2.5 API page.

What is Seedance 2.5?

Seedance 2.5 is the newest model in ByteDance’s Seedance video line, part of the company’s broader Seed family of foundation models. The Seedance models are built around motion quality, director-level camera and scene control, and consistency held from one shot to the next — the things production work actually needs.

The jump from Seedance 2.0 comes down to two numbers:

1. Clip length, which reportedly moves up to 30 seconds in a single pass.
2. Reference inputs, which reportedly move up to 50 per request.

Native 30-second generation

Seedance 2.5 reportedly generates a full 30 seconds of video in one pass. That is the change most people noticed first, because clip length has been the main ceiling on AI video for a while now.

Today’s models tend to top out at a handful of seconds — Seedance 2.0 on WaveSpeedAI goes up to 15 seconds — so anything longer gets assembled from several short clips joined end to end, or stretched with extend endpoints like Seedance 2.0 Video Extend. Splicing works, but the seams can show: lighting drifts, motion resets, characters wobble between segments.

A native 30-second pass means a scene can play out without visible cuts. For advertisers, that is a complete 30-second spot directed from start to finish in one generation.

Up to 50 multimodal references

The other change is on the input side: up to 50 references per request, according to media reports. A reference is anything you give the model to steer the result — still images, video clips, audio, or specific frames you want it to land on — and reports say they can be mixed freely in one generation.

The practical effect: you can describe a shot the way you would brief a film crew, with the character, the setting, the camera move, and the mood all specified at once. The more real material the model has to work from, the closer the first result lands to what you actually wanted — and the fewer attempts you burn getting something usable.

For teams with an existing asset library, this is the real win. Your locked-down brand look, your product shots, your reference footage all travel straight into the generation instead of being approximated from a text prompt.

Will Seedance 2.5 generate in 4K?

Reports say yes — those 30-second clips can reportedly render in 4K. Seedance 2.0 on WaveSpeedAI already outputs up to 4K, so this would carry the same ultra-HD ceiling into much longer clips.

Seedance 2.5 vs Seedance 2.0: how they compare

Seedance 2.0 is live on WaveSpeedAI today, so it is the clearest reference point for what 2.5 changes. The 2.0 figures below come from the live WaveSpeedAI endpoints; the 2.5 figures are reported and unconfirmed.

Detail	Seedance 2.0	Seedance 2.5 (reported)
Status	Live on WaveSpeedAI	Announced, coming soon
Max clip length	Up to 15 seconds	Up to 30 seconds
Resolution	480p / 720p / 1080p / 4K	4K
Inputs	Text, image, audio, video	Text, image, audio, video
Reference inputs	Multiple mixed references	Up to 50 multimodal (reported)
Native audio	Yes	Expected, not confirmed

When will Seedance 2.5 be released?

Media coverage points to a launch around July 2026, but ByteDance has not put an official date on it. Until it does, treat the timing as a moving target.

What we can say: Seedance 2.5 will be available on WaveSpeedAI when it launches. New ByteDance models tend to reach WaveSpeedAI fast — the full Seedance 2.0 family (text-to-video, image-to-video, video edit, video extend, plus Turbo and Mini tiers) is already running on the platform today.

Why run Seedance 2.5 on WaveSpeedAI?

For a model like Seedance 2.5, a few platform details matter more than usual:

• Heavy renders, no cold starts. A 30-second 4K clip is a heavy generation. WaveSpeedAI’s inference infrastructure is built for exactly this kind of load, with fast queuing and capacity that scales with demand.
• One API for a multi-input workflow. A 50-reference request mixing images, video, and audio stays simple when every model runs behind one API and one key. No separate integrations to wire up.
• Pay-as-you-go pricing. Billing is per generation with no subscription, so heavy bursts of 4K rendering do not leave you paying for idle capacity afterwards.
• Zero migration. If you are already running Seedance 2.0 on WaveSpeedAI, 2.5 will arrive on the same setup — same API pattern, same key, nothing new to build.

Get ready today with Seedance 2.0

The best way to prepare for Seedance 2.5 is to have your pipeline running on Seedance 2.0 now. The integration is a few lines:

import wavespeed

output = wavespeed.run(
    "bytedance/seedance-2.0/text-to-video",
    {
        "prompt": "A cinematic 30-second product reveal: slow dolly-in on a smartwatch on a marble desk, morning light, soft depth of field",
        "duration": 15,
        "resolution": "1080p",
    },
)

print(output["outputs"][0])  # Output video URL

Image-to-video works the same way — pass an "image" key with your reference:

import wavespeed

output = wavespeed.run(
    "bytedance/seedance-2.0/image-to-video",
    {
        "image": "https://your-cdn.com/product-shot.png",
        "prompt": "Slow push-in with gentle parallax, studio lighting holds steady",
        "duration": 10,
    },
)

print(output["outputs"][0])

When Seedance 2.5 goes live, switching will be a model-ID change, not a rebuild.

Frequently asked questions about Seedance 2.5

What is Seedance 2.5?

Seedance 2.5 is ByteDance’s next-generation video model, the follow-up to Seedance 2.0 in the company’s Seed family. The standout reported changes are 30-second clips in a single pass and up to 50 mixed references per request.

When will Seedance 2.5 be released?

A launch is reported for around July 2026, though ByteDance has not confirmed an official date. It will be available on WaveSpeedAI when it ships — watch the Seedance 2.5 API page.

What is the maximum Seedance 2.5 clip length?

Coverage points to 30 seconds generated in one pass. That stands out because most current models make shorter clips and reach length by joining segments, which can show at the seams.

How many references can Seedance 2.5 use?

Reports put it at up to 50, with images, video, and audio allowed in the same request. The exact limits are worth confirming against ByteDance’s eventual documentation.

How is Seedance 2.5 different from Seedance 2.0?

Seedance 2.0 is live now on WaveSpeedAI with clips up to 15 seconds and output up to 4K. Seedance 2.5 reportedly pushes clip length to 30 seconds and reference inputs to 50, with 4K output in the same model.

How much will Seedance 2.5 cost on WaveSpeedAI?

No pricing has been announced yet. WaveSpeedAI bills video models pay-as-you-go with no subscription, and Seedance 2.5 will follow the same model. For a baseline today, see Seedance 2.0 pricing.

---

Seedance 2.5 is coming soon to WaveSpeedAI. Bookmark the Seedance 2.5 API page to be ready on day one — and start building now with the live Seedance 2.0 API.