Companies / ByteDance
ByteDance
Also known as: 字节跳动, Doubao, 豆包
ByteDance is a Chinese AI company headquartered in No. 1 Building, Dazhongsi Plaza, No. 18A North Third Ring Road West, Haidian District, Beijing, China, known for the doubao-seed-2-1-pro, doubao-seed-evolving, doubao-seed-2-1-turbo models.
| Key fact | Value |
|---|---|
| Founded | Unknown |
| Headquarters | No. 1 Building, Dazhongsi Plaza, No. 18A North Third Ring Road West, Haidian District, Beijing, China |
| Funding | Privately held; funding rounds not officially disclosed as of 2026-09-22. |
| Official website | https://www.bytedance.com/ |
| Documentation | https://docs.volcengine.com/docs/ark |
| AI products | Doubao (consumer AI assistant), Volcengine Ark (AI API platform), Seedance (video generation), Seedream (image generation) |
| Cloud distribution | Volcengine Ark (cn-beijing) |
Major releases
| Release | Date | Type |
|---|---|---|
| Doubao Seed 2.1 family | 2026-06-23 | model_release |
| Doubao Seed 2.1 Pro with 1M context (260915) | 2026-09 | model_release |
| Doubao Seed Evolving (rolling weekly-updated model) | 2026-06 | model_release |
Open-source projects
- VeOmni (ByteDance-Seed, scaling multimodal model training)
- Triton-distributed (ByteDance-Seed)
ByteDance is a Chinese technology company and the provider of the Doubao (豆包) AI assistant and the Doubao Seed family of foundation models. Its API platform is Volcengine Ark (火山方舟), operated by Beijing Volcano Engine Technology Co., Ltd. The ByteDance Seed research team publishes the official Seed blog and maintains the ByteDance-Seed GitHub organization, which hosts open-source training tooling such as VeOmni and Triton-distributed (but no Doubao LLM weights).
Doubao Seed models are closed-API: they are served through Ark in the cn-beijing region. The current flagship text models are Doubao Seed 2.1 Pro (1M-token context as of September 2026), Doubao Seed Evolving (a rolling model updated at least weekly for agent and coding use), and Doubao Seed 2.1 Turbo. Video generation (Seedance) and image generation (Seedream) are separate dedicated models.
Models
Related entities
Sources
Data interpretation
| Field | Value | Evidence type |
|---|---|---|
| Headquarters | No. 1 Building, Dazhongsi Plaza, No. 18A North Third Ring Road West, Haidian District, Beijing, China | Official |
| Foundation models | doubao-seed-2-1-pro, doubao-seed-evolving, doubao-seed-2-1-turbo | Official |
| Major releases | 3 entries | Official |
| Entity relations | 3 models, 0 agents, 0 APIs | China AI Hub analysis |
Evidence types: Official = vendor documentation or official pages. China AI Hub analysis = derived from the database itself. See the sourcing policy.
Where is ByteDance based?
No. 1 Building, Dazhongsi Plaza, No. 18A North Third Ring Road West, Haidian District, Beijing, China.
Which foundation models does ByteDance offer?
doubao-seed-2-1-pro, doubao-seed-evolving, doubao-seed-2-1-turbo.
Does ByteDance provide an API?
No API is currently listed for ByteDance in our database.
What open-source projects does ByteDance maintain?
VeOmni (ByteDance-Seed, scaling multimodal model training), Triton-distributed (ByteDance-Seed).
Where does China AI Hub get its ByteDance data?
From 5 sources, last verified 2026-09-20.