What Is Seedance 2.0 and How to Use It (2026 Guide)
What Seedance 2.0 is, which sites are actually official, every platform where you can use it today, and how to call the API. Verified July 2026, no guesswork.
Seedance 2.0 is ByteDance's flagship AI video model: text, images, audio, or clips in, up to 15 seconds of 480p-to-4K video with synchronized sound out[1]. Since its February 12, 2026 launch[2] it has topped the major blind-test leaderboards[3][4], and "how do I actually use this" became one of the most-searched questions in AI video.
Fair question, because the access story is genuinely confusing. The model goes by three names across ByteDance's own properties, there is no single "seedance.com," and the top search results for the name are squatter sites that admit in their own terms of service they have nothing to do with ByteDance. This guide sorts it out: what the model is, which sites are real, every place you can use Seedance 2.0 today, and what calling the API looks like.
TL;DR
- Seedance 2.0 is ByteDance's video generation model: 480p to 4K output, 4 to 15 seconds per clip, references of up to 9 images plus 3 videos plus 3 audio clips, with audio generated alongside the video[1].
- It leads the leaderboards: #1 on Artificial Analysis text-to-video (Elo 1,222, the largest lead on the board) and #1 on LMArena image-to-video (1,474 on ~82,000 votes)[3][4].
- There is no single official website. ByteDance splits it three ways: seed.bytedance.com for model info, Dreamina (international) and Jimeng (China) for consumer creation, BytePlus and Volcano Engine for the enterprise API[2].
- The top search results are not ByteDance. Sites like seedance2.ai and seedance.ai sell subscriptions while disclaiming any affiliation in their own fine print. Check before you pay.
- You cannot run it locally, and ChatGPT does not have it. The weights are closed, and OpenAI has never offered Seedance[5].
- API access runs from $0.0400/s on reAPI's Seedance 2.0 lineup, with per-second billing and no subscription[6].
What Seedance 2.0 actually is
Seedance 2.0 is the second-generation video model from ByteDance's Seed research group, released on February 12, 2026[2]. The pitch that separates it from most rivals is multimodal referencing: alongside a text prompt you can attach up to 9 images, 3 video clips, and 3 audio tracks in one generation, and the model composes them into a single scene[1]. That is how creators keep the same character across a whole series of clips, clone a camera move from a reference video, or drive a performance from a song.
The published spec sheet, per ByteDance's own model documentation: output from 480p up to 4K (the 4K tier renders in 10-bit H.265), durations of 4 to 15 seconds at 24fps, and audio generated jointly with the picture rather than bolted on afterward[1]. Lighter Fast and Mini variants trade the top resolutions for lower cost.
The blind-test numbers back the hype up. On Artificial Analysis, Seedance 2.0 ranks #1 in text-to-video at Elo 1,222, and the 71-point gap to second place is the largest lead anywhere on that board; it holds #1 in image-to-video too[3]. On LMArena it is #1 in image-to-video at 1,474 across roughly 82,000 votes, ahead of Veo 3.1 at 1,391[4]. Five months after launch, it is still the model everyone else is measured against, which is also why ByteDance has already announced a successor, Seedance 2.5, for July.
One asterisk on availability: after copyright disputes with Hollywood studios in March 2026, ByteDance paused the global rollout and resumed it everywhere except the United States[7]. US access remains restricted on ByteDance's own consumer surfaces; third-party platforms have varying policies.
The real official websites (and the fake ones)
The single most-searched Seedance question after "what is it" is some version of "what is the real official website." Here is the verified answer.
ByteDance splits the model across three official doors, and the model info page's own buttons prove the mapping: the "Try Now" button on seed.bytedance.com routes to Dreamina, and "Get API" routes to BytePlus[2].
| You want to | Official destination |
|---|---|
| Read about the model | seed.bytedance.com (ByteDance Seed research site)[2] |
| Make videos as a consumer | Dreamina (dreamina.capcut.com, international) or Jimeng (China)[8] |
| Call it programmatically | BytePlus ModelArk (global) or Volcano Engine (China)[1][9] |
Adding to the confusion, ByteDance itself uses three names for the same model: Volcano Engine calls it Doubao Seedance 2.0, while BytePlus and Dreamina call it Dreamina Seedance 2.0. Same model, different storefronts.
Now the fake ones. When I checked in early July 2026, the #1 organic result for "seedance," "seedance 2.0," and "seedance ai" was in every case a lookalike domain, with ByteDance's own pages ranking second to fifth. A few receipts from the squatters' own pages: seedance2.ai and seedance.ai both state in their terms that they are "not affiliated with ByteDance" while charging $15 to $100 a month; one operator runs multiple domains with word-for-word identical pricing copy; another is registered to an anonymous Wyoming shell company; one left "[Your Country or State]" template placeholders in its legal page. None of these sites is ByteDance, and paying them buys you, at best, resold generations at a markup.
The safe rule: if the domain is not bytedance.com, capcut.com, byteplus.com, volcengine.com, or a platform you already know by reputation, assume it is not official.
Everywhere you can use Seedance 2.0 today
I verified each of these on July 3, 2026 by visiting the platform directly. Access modes vary a lot, and two popular assumptions are wrong: OpenArt does not carry the 2.0 model (its newest Seedance is 1.5 Pro), and ChatGPT has never offered Seedance at all[5].
| Platform | Seedance 2.0? | Access model |
|---|---|---|
| Dreamina[8] | Yes (2.0 + Fast) | Free login + credit system |
| Jimeng (China) | Yes | Membership + credits |
| BytePlus ModelArk[9] | Yes (2.0 / Fast / Mini) | Enterprise API, token billing |
| Higgsfield[10] | Yes (2.0 + Fast) | Subscription; the $19 Starter tier does NOT include 2.0, full access starts at Plus (~$47/mo annual) |
| Krea[11] | Yes (2.0 / Fast / Mini) | Subscription compute units; Basic $9/mo covers roughly 20 videos |
| Pollo.ai[12] | Yes | Subscription from $15/mo (Lite, watermarked) |
| fal.ai[13] | Yes | API, per-second: $0.30/s at 720p, $0.68/s at 1080p |
| Replicate[14] | Yes (official listing) | API, per-second: $0.18/s at 720p, $0.45/s at 1080p, $1/s at 4K |
| reAPI[6] | Yes (2.0 / Fast / Mini) | API, per-second from $0.0400/s, plus Mini from $0.03/s; no subscription |
| OpenArt | No (tops out at Seedance 1.5 Pro) | — |
| ChatGPT | Never had it[5] | — |
Two other questions from the search data worth answering flat: you cannot install Seedance 2.0 locally, because ByteDance has not released the weights; anything labeled "Seedance local install" is a different model wearing the name. And the GPT Store listings named "Seedance" are user-made wrappers, not the model. OpenAI is in fact winding its own video product down; Sora's web app closed in April 2026 and its API retires in September[5].
How to use Seedance 2.0, both routes
The consumer route. Sign into Dreamina with a free account, pick Seedance 2.0 or 2.0 Fast in the video tool, write a prompt, optionally attach reference images, and spend credits per generation[8]. This is the right door for trying the model, with two caveats: daily free credits are modest, and Dreamina blocks real-person faces on 2.0 uploads.
The API route. For anything repeatable, whether a content pipeline, an app feature, or bulk e-commerce clips, you want per-second API billing rather than subscription credits. On reAPI the call is one POST:
curl -X POST https://reapi.ai/api/v1/videos/generations \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "doubao-seedance-2.0",
"prompt": "A kitten yawning at the camera",
"size": "16:9",
"resolution": "720p",
"duration": 5
}'The request returns a task id; poll GET /api/v1/tasks/:id until the clip is ready. Reference images, video clips, and audio tracks attach as public URLs, up to the model's 9 image / 3 video / 3 audio caps[1], and reference-mode generations actually bill lower than pure text-to-video. Full parameter docs live at reapi.ai/docs/seedance-2-0.
Which route is yours: if you are exploring what the model can do, Dreamina costs nothing to try. The moment you generate on a schedule or inside a product, per-second API pricing beats subscription math, and switching tiers (Standard, Fast, Mini) is a one-word change to the model string.
What Seedance 2.0 costs, in one paragraph
Official API pricing is token-based and lands around $3.89 for a 5-second 4K clip on BytePlus[9]. Per-second platforms are easier to reason about: Replicate charges $0.18/s at 720p[14], fal.ai $0.30/s[13], and reAPI runs Seedance 2.0 from $0.0400/s to $0.4048/s depending on tier, resolution, and reference mode, with the Mini tier from $0.03/s[6]. Subscriptions (Higgsfield, Krea, Pollo) make sense mainly if you also use those platforms' editors. The live pricing table has current per-second rates; a full platform-by-platform cost breakdown is its own article.
FAQ
What is Seedance 2.0 in one sentence?
ByteDance's flagship video generation model: prompts plus optional image, video, and audio references in; up to 15 seconds of 480p-to-4K video with synchronized audio out[1].
What is the real Seedance 2.0 official website?
There isn't one single site. Model information lives at seed.bytedance.com, consumer creation at Dreamina (dreamina.capcut.com) internationally and Jimeng in China, and the official API at BytePlus and Volcano Engine[2]. Domains built around the "seedance" name, like seedance2.ai and seedance.ai, are unaffiliated resellers by their own terms of service.
Is Seedance 2.0 free to use?
Trying it is free: Dreamina gives logged-in users daily credits[8], and several API platforms include signup credit. Sustained free usage is not a thing; the "unlimited Seedance" annual deals some platforms advertised have been quietly withdrawn or capped.
Can I install Seedance 2.0 locally?
No. The weights are closed and ByteDance offers no download. Local "Seedance" installs are other open models relabeled. If you need self-hosted video generation, look at open-weight models; if you need Seedance itself, the API is the only programmatic route.
Does ChatGPT have Seedance 2.0?
No, and it never has[5]. GPT Store items named "Seedance" are user-created wrappers. OpenAI's own video model, Sora, is being retired through 2026.
Why can't I use Seedance 2.0 in the US?
After copyright disputes with studios, ByteDance paused and then resumed the global rollout everywhere except the United States[7]. Some third-party platforms serve US users anyway under their own arrangements; ByteDance's consumer apps do not.
How long can Seedance 2.0 videos be?
4 to 15 seconds per generation[1]. The announced successor, Seedance 2.5, claims 30-second single-pass generations; see our Seedance 2.5 pre-launch guide.
When did Seedance 2.0 come out?
February 12, 2026, announced by ByteDance Seed[2]. Global consumer rollout followed in late March, and API access opened in mid-April 2026.
Picking your door into Seedance 2.0
Sorted by intent: read about the model at ByteDance Seed, play with it free on Dreamina, and build on it through a per-second API. Skip anything with "seedance" in the domain that is not ByteDance's, skip the GPT Store clones, and do not wait for a local version that is not coming. If you are heading down the API route, Seedance 2.0 on reAPI runs the full Standard, Fast, and Mini lineup from $0.03/s with signup credits to test on, so the distance from reading this to your first generated clip with Seedance 2.0 is about one curl command.
References
- Volcano Engine (ByteDance). Doubao Seedance 2.0 — model specifications (resolutions, durations, reference limits). Retrieved July 2026 from volcengine.com/docs/82379/1330310
- ByteDance Seed. Seedance 2.0 — official model page and launch announcement. Retrieved July 2026 from seed.bytedance.com/en/seedance2_0
- Artificial Analysis. Text to Video and Image to Video Leaderboards. Retrieved July 2026 from artificialanalysis.ai/video/leaderboard/text-to-video
- LMArena. Image to Video Leaderboard. Retrieved July 2026 from arena.ai/leaderboard/image-to-video
- OpenAI. Sora sunset notice (help center). Retrieved July 2026 from help.openai.com/en/articles/20001152
- reAPI. Seedance 2.0 — model page and live pricing. Retrieved July 2026 from reapi.ai/models/seedance-2-0
- Reuters. ByteDance suspends launch of video AI model after copyright disputes. Retrieved July 2026 from reuters.com/technology/bytedance-suspends-launch-video-ai-model
- Dreamina (CapCut). AI video generation — Seedance model picker. Retrieved July 2026 from dreamina.capcut.com/ai-tool/home
- BytePlus. ModelArk — Seedance 2.0 documentation and pricing. Retrieved July 2026 from docs.byteplus.com/en/docs/ModelArk/1330310
- Higgsfield. Pricing. Retrieved July 2026 from higgsfield.ai/pricing
- Krea. Video generation. Retrieved July 2026 from krea.ai/video
- Pollo.ai. Seedance 2.0. Retrieved July 2026 from pollo.ai/m/seedance-2-0
- fal.ai. Seedance 2.0 — Text to Video. Retrieved July 2026 from fal.ai/models/bytedance/seedance-2.0/text-to-video
- Replicate. bytedance/seedance-2.0. Retrieved July 2026 from replicate.com/bytedance/seedance-2.0
Further reading
- reAPI. Seedance 2.5: What We Know Before the Public Launch. reapi.ai/blog/seedance-2-5-what-we-know-2026
- reAPI. Veo 3.1 vs Seedance 2.0. reapi.ai/blog/veo-3-1-vs-seedance-2-0-2026
- reAPI. Seedance 2.0 API documentation. reapi.ai/docs/seedance-2-0
Автор
Категории
Ещё статьи
What Is Claude Opus 4.8? Anthropic's New Model Explained
Claude Opus 4.8 is Anthropic's most capable model for reasoning and agentic coding. Here is what's new, its benchmarks, pricing, and how to access it.
Gemini Omni vs Veo 3.1: Should You Migrate in May 2026?
Gemini Omni vs Veo 3.1 in May 2026: Google says Omni replaces Veo in the Gemini app, not in the API. Five-channel mapping, code diff, where each wins.
Best Replicate Alternatives in 2026: 5 Options Compared
Looking for Replicate alternatives in 2026? Compare fal.ai, Together AI, RunPod, Hugging Face, and reAPI on model range, pricing, speed, and API design.