Seedance FAQ

Seedance is ByteDance's family of AI video generation models — it turns text prompts and images into short video clips. The line started with Seedance 1.0 in June 2025; Seedance 2.0 (February 2026) is today's widely available version with native audio and lip sync, and Seedance 2.5 was announced in June 2026.

Seedance is developed and owned by ByteDance, the company behind TikTok, CapCut, Doubao and Dreamina. ByteDance distributes it through its own platforms (Dreamina, CapCut, Volcano Engine, BytePlus) and through licensed API partners such as fal and Replicate.

Seedance 2.0 is ByteDance's current-generation AI video model, released in February 2026. It's a unified multimodal generator — it takes text, image, audio, and video inputs and produces short clips with native audio, cinematic motion, and strong prompt control.

Cinematic establishing shots, product turntables, anime scenes, character motion, and more — anything you can describe as a short video clip.

Text-to-video generates a scene from your description. Image-to-video animates a still image you provide, preserving its composition while adding motion.

Seedance 1.0 is ByteDance's first publicly released Seedance AI video model, launched on June 11, 2025 at the Volcano Engine Spring FORCE conference. It generates text-to-video and image-to-video clips of roughly 2–12 seconds at up to 1080p/24fps, with multi-shot storytelling and first/last-frame control.

Seedance 1.0 generates silent clips (no audio) of 2–12 seconds at up to 1080p. Seedance 2.0 (February 2026) is a unified multimodal model: it accepts text, image, audio and video references, generates native audio including lip sync, produces ~15-second clips, and a later update added native 4K. For most work, 2.0 is the better choice.

Only "Seedance 1.5 pro", released December 16, 2025 — there is no separate non-pro Seedance 1.5. Its headline feature was native synchronized audio and video, which set the direction for Seedance 2.0.

Seedance 2.0 was released on February 12, 2026. The line began with Seedance 1.0 (June 11, 2025), followed by Seedance 1.5 pro (December 16, 2025), and then 2.0 — which remains the version most people use today.

Seedance 2.5 was officially announced on June 23, 2026 at the Volcano Engine FORCE conference and remains in closed (enterprise) beta as of mid-July 2026. Third-party reports pointed to a mid-July API opening, but no public API, model listing, or pricing has materialized yet, and ByteDance has not confirmed an exact public launch date — so it is not generally available.

ByteDance announced three headline upgrades: native 30-second clips in a single generation, up to 50 reference inputs in one prompt, and localized controllable editing — changing part of a shot without regenerating the whole clip. Note that native 4K belongs to a Seedance 2.0 update, not 2.5.

Seedance 2.5, announced June 23, 2026, is the newest version — but it's still in closed enterprise beta. The latest version you can actually use today is Seedance 2.0, available through Dreamina, CapCut, the APIs, and ponpon.

It depends on how you access it. Platforms that offer Seedance set their own pricing — often a mix of free credits, pay-as-you-go, and subscriptions. Our prompts are free to copy; check your provider for current generation rates.

There's no single price — it depends on the platform. Consumer apps like Dreamina and CapCut sell credit plans with limited free allowances, while APIs bill per output: ByteDance's Volcano Engine / BytePlus ModelArk charges per token, and partners like fal and Replicate charge per second of video. Prices vary by resolution and change often, so check your platform's current rate card.

A few genuinely free paths exist: Dreamina's daily free credits (outputs carry a watermark), BytePlus ModelArk's free token grant for new API accounts, and the Doubao app in China. Free tiers are fine for testing prompts — for regular or watermark-free generation you'll need a paid plan.

Yes. Seedance 2.0 is generally available, and creators with early 2.5 access are already sharing working prompts — you can copy them from our library and run Seedance on ponpon today.

Yes. Outside mainland China you can reach Seedance through international platforms — Dreamina's global app, CapCut, and API partners such as fal and Replicate — as well as ponpon. ByteDance's Jimeng (即梦) app is China-only and needs a mainland phone number, but the others don't.

Yes — mainly through APIs. ByteDance's Volcano Engine / BytePlus ModelArk still serves doubao-seedance-1.0-pro, and partners like fal expose Seedance v1 endpoints. Consumer apps such as Dreamina have moved to newer versions, so for hands-on use most people are on Seedance 2.0 today.

Yes. ByteDance serves Seedance through Volcano Engine Ark and BytePlus ModelArk (the doubao-seedance-* models, billed per token), and third parties like fal and Replicate offer per-second Seedance endpoints.

No. Seedance is a proprietary ByteDance model with no public weights, so you can't run it locally — only through official or third-party APIs and apps.

Write a prompt that names the subject, camera movement, lighting, and style, pick an aspect ratio, and generate. Use the generator on the Seedance 2.0 page to open Seedance on ponpon with your prompt pre-filled — or copy any prompt from our library.

Lip sync is built into Seedance 2.0's generation: write the dialogue into your prompt — who's speaking, their tone, the line — and the model generates voice and matching mouth movement in the same pass. In Dreamina's Multiframes mode you can also upload audio and reference it with @audio1. If you want to make a still photo speak from an audio file instead, that's OmniHuman, ByteDance's separate talking-avatar tool inside Dreamina.

A clear subject, an explicit camera movement, defined lighting/mood, and a style keyword. Stacking those four consistently produces usable clips.

Add explicit motion — a camera move, wind, particles, or subject action. Descriptions without movement tend to produce static-feeling clips.

Describe only subtle motion (a gentle head turn, slight breeze) and add 'preserve identity' to your prompt. Start from a sharp, well-lit source image.

Seedance's strengths are native audio, strong prompt adherence, and cinematic motion. Kling and Veo 3 trade off differently on clip length, resolution, audio, and access, while OpenAI's Sora was discontinued in April 2026 (its API shuts down that September) — we compare them without relying on unverified benchmark scores.

Yes. ponpon runs Seedance for both text-to-video and image-to-video with no setup or API keys — copy a prompt from our library or type your own on the Seedance 2.0 generator, and it opens on ponpon pre-filled.