Doubao Seed

weights

closed

data

closed

code

closed

license

Proprietary(Doubao app + Volcano Engine API only

No weights released, served only through the Doubao app and Volcano Engine API. Fully closed.

• https://seed.bytedance.com/en/blog/seed2-0-%E6%AD%A3%E5%BC%8F%E5%8F%91%E5%B8%83 recorded 2026-09-29

  ByteDance launch post lists only hosted access: "the model API for the Seed2.0 full series is available on Volcano Engine". Searched the body for open source, open weight, license, download and weights: none.

• https://technode.com/2026/02/14/bytedance-releases-doubao-seed-2-0-positions-pro-model-against-gpt-5-2-and-gemini-3-pro/ recorded 2026-09-29

  The article states "The model is now available on the Doubao app and through Volcano Engine API services." It mentions no weight release, training corpus, training code or license. Searched the body for open source, open-weight, self-host, license, download and weights: none.

Adoption is read from the Doubao app's reported weekly active users, because Seed 2.0 powers that app and no standalone usage figure exists for the model, which is also sold through Volcano Engine. The figure reaches this entry through press coverage of a Quest Mobile report rather than the report itself, and the separate Doubao entry reports monthly users of a similar order of magnitude.

• https://technode.com/2026/02/14/bytedance-releases-doubao-seed-2-0-positions-pro-model-against-gpt-5-2-and-gemini-3-pro/ recorded 2026-09-29

  TechNode launch coverage: "The model is now available on the Doubao app and through Volcano Engine API services", linking this model to the app whose user figure is read. It carries no user count.

• https://www.caixinglobal.com/2026-02-15/bytedance-unveils-doubao-20-ai-model-to-tackle-complex-tasks-102414865.html recorded 2026-09-29

  Caixin Global on the Seed 2.0 launch: "Doubao led with 155 million weekly active users in December 2025" per Quest Mobile, ahead of DeepSeek; Tencent Yuanbao had "20.8 million weekly active users".

This score reads Seed2.1 Pro, ByteDance's current flagship. Artificial Analysis does not track it, so the evidence is ByteDance's own table, which puts it level with Gemini 3.1 Pro, Claude Opus 4.7 and GPT-5.5: ahead on agentic work, behind on knowledge benchmarks. Those peers now sit in the upper middle of the Artificial Analysis index, well behind the leading group, which places Seed2.1 Pro beside Gemini Pro rather than at the frontier.

• https://seed.bytedance.com/en/models recorded 2026-10-08

  ByteDance Seed models page lists "Seed2.1 A next-generation agent for real-world productivity" as the current language model, with no Seed2.0 entry.

• https://seed.bytedance.com/en/seed2_1 recorded 2026-10-08

  Seed2.1 page, "Evaluation Results" table with columns "Seed2.1 Pro Seed2.1 Turbo Claude Opus 4.7 GPT-5.5 Gemini 3.1 Pro": "SuperGPQA Knowledge 70.8 67.4 68.5 72.7 76.6", "Agent Startup Bench High-Economic-Value 68.8 54.0 62.3 68.1 45.7", "Terminal Bench 2.1 Terminal Usage 71.0 67.6 71.7 73.8 70.7".