Seedance
Category: Generative Media (Image / Video)
A next-generation multimodal AI video generation platform by ByteDance that creates cinematic-quality, multi-shot videos with native synchronized audio from text, image, and audio inputs. Seedance was founded in 2023. The company is led by Wu Yonghui. Based in Beijing, China. Team size: 1,500. Total funding raised: Undisclosed (Corporate Funded). Latest round: Corporate Round. Key investors include ByteDance.
- Founded
- 2023
- Headquarters
- Beijing, China
- Team size
- 1,500
- Total funding
- Undisclosed (Corporate Funded)
Value proposition
Delivers Hollywood-grade video production capabilities by automating complex multi-shot narrative coherence and high-fidelity audio-visual synchronization, significantly reducing traditional production costs and timelines.
Products and solutions
Seedance 2.5 (flagship: 30s one-take storytelling, multimodal reference, timestamp editing), Seedance 2.0 (unified multimodal audio-video joint generation; text/image/audio/video inputs), Seedance 1.5 pro (native joint audio-video foundation model), Seedance 1.0 (multi-shot text/image-to-video), Jimeng AI / Dreamina (consumer distribution), CapCut Dreamina Seedance integration, Seedance API via Volcengine / BytePlus ModelArk
Unique value
Adopts a 'unified multimodal audio-video joint generation architecture' that allows the model to process and generate video and audio simultaneously, ensuring perfect lip-sync and ambient sound alignment without post-production.
Target customer
Professional filmmakers, advertising agencies, social media content creators, and enterprise marketing teams.
Industries served
Film & Entertainment, Advertising & Digital Marketing, Social Media & Content Creation, E-commerce, Gaming & Animation
Technology advantage
Features a proprietary physics-aware training module that accurately simulates real-world dynamics (fluidity, smoke, gravity) and maintains 'Multi-Shot Narrative' coherence, allowing users to generate consistent characters and settings across different camera angles in a single project.
How they differentiate
Utilizes a unified multimodal audio-video joint generation architecture that produces native synchronized audio and video simultaneously, alongside a 'Multi-Shot Narrative' engine that maintains character and environmental consistency across different camera angles.
Main competitors
OpenAI (Sora / Sora 2), Google (Veo), Kling AI (Kuaishou), Runway, Luma AI (Dream Machine)
Key partnerships
Volcengine / BytePlus ModelArk (enterprise API and cloud), Jimeng AI / 即梦 (China consumer distribution), Dreamina (global creative platform), CapCut (phased global Seedance rollout from Mar 2026 with IP/face safeguards and C2PA-style watermarking), Doubao Pro (in-app video generation), Legal note: Ongoing Hollywood IP scrutiny after Feb 2026 MPA/Disney/Paramount C&Ds; global access resumed with content filters
Notable customers
TikTok / Douyin creator network, Jimeng AI (即梦) and Dreamina users, CapCut creators (phased global markets), Volcengine / BytePlus enterprise API customers, Advertising and marketing teams via Pippit
Major milestones
ByteDance Seed team established early 2023 (pre-Wu), Wu Yonghui joins as Seed foundational research head (Feb 2025), reporting to CEO Liang Rubo, Seedance 1.0 multi-shot video model launch (2025), Seedance 1.5 pro native audio-video joint generation (Dec 2025), Seedance 2.0 official launch with unified multimodal audio-video architecture (Feb 12, 2026), MPA, Disney, Paramount and other studios issue C&Ds over alleged copyright infringement (Feb 2026), Global rollout briefly constrained for IP safeguards; CapCut/Dreamina phased international expansion from late Mar 2026 with face/IP blocks and watermarking, Seedance 2.5 launch: up to 30s single-pass storytelling, up to 30 image / 10 video / 10 audio refs, timestamp editing (Jul 31, 2026)
Growth metrics
Seed team reported at thousand-plus scale under ByteDance; Seedance distributed via Jimeng/Dreamina, CapCut (phased global markets from Mar 2026), Doubao, and BytePlus/Volcengine API; Seedance 2.5 rolling out on Jimeng and Doubao Pro (Jul 31, 2026) with API forthcoming.
Market positioning
High-end cinematic AI video generation for professional filmmakers and enterprise marketing, positioned as a direct 'Sora-killer' with integrated production tools.
Geographic focus
Global (with primary R&D in China and major market expansion in North America and Europe via the TikTok/CapCut ecosystem).
Patents and IP
Proprietary architectures for 'Temporal Consistency in Diffusion Models' and 'Multimodal Joint Embedding for Video-Audio Synthesis' (specific patent IDs are typically held under ByteDance Ltd. corporate filings).
About Wu Yonghui
Dr. Wu Yonghui is a world-renowned AI scientist and former Google Fellow, previously VP of Research at Google DeepMind. He is best known as a lead author of Google's Neural Machine Translation (GNMT) system and a principal contributor to Gemini and foundational deep learning architectures during ~17 years at Google. He joined ByteDance in February 2025 as Head of Foundational Research for the Seed team (Head of Research, ByteDance Seed), reporting to CEO Liang Rubo; Zhu Wenjia continues to lead Seed model applications.
Latest news about Seedance
- ByteDance has officially unveiled Doubao 2.0, integrating its groundbreaking Seedance 2.0 video engine into China's most popular AI application. The new Doubao-Seed-2.0 Pro model now benchmarks alongs
- ByteDance has unveiled Seedance 2.0, a multimodal AI video model that processes text, images, audio, and video simultaneously. The system achieves a 90% usable output rate on the first try, shifting A
More Generative Media (Image / Video) companies
Official website: https://seed.bytedance.com/
