Dola Seed 2.1 Turbo
Language modelCard summarizedSeed2.1 Turbo is designed for efficient, high-throughput product scenarios
Source: Seed2.1 Model Card: Agentic Intelligence for Productivity (73 pp)
Use it via API
Where to call it
- BytePlus ModelArk
dola-seed-2-1-turbo-260628
Supports
- Structured outputs
- Streaming
- Reasoning control
- Tool calling
- Batch API
- Image input
- Prompt caching
Good for
- High-throughput agent products: office documents, spreadsheets, reports (Workspace Bench 54.7)
- Tool use through MCP (MCP-Atlas 80.3, above Claude Opus 4.7 and Gemini 3.1 Pro)
- Search agents (BrowseComp with search 84.9)
Overview
Seed 2.1 Turbo
Seed2.1 Turbo is designed for efficient, high-throughput product scenarios
- Cost
- $0.50 in
- Context
- 256K
- Input / output
- $0.50 / $2.50
- Max output
- 256K
- Knows up to
- —
- Released
- 23 Jun 2026
Seed2.1 comes in two sizes: Turbo for efficient, high-throughput product scenarios and Pro for stronger reasoning and agentic execution. ByteDance developed and evaluated it around real product use (Claude Code, Trae, OpenClaw-style agents). Turbo is close to the frontier on general-agent and tool-use tasks (Workspace Bench 54.7, MCP-Atlas 80.3, BrowseComp 84.9) but trails GPT-5.5 and Claude Opus 4.7 on long-horizon coding and abstract reasoning. §1, Tables 1–2, 5
dola-seed-2-1-turbo-260628- Input / output per 1M
- $0.50 / $2.50
- Context window
- 256K
- Max output
- 256K
- Knows up to
- —
- Reasoning
- Deep reasoning on by default; can be disabled; reasoning_effort supported
- Weights
- Closed
About this card. A model card is the maker's own account of a model: what it's for, how it was trained and tested, and where it falls short. This page condenses ByteDance Seed's Seed2.1 Model Card: Agentic Intelligence for Productivity. Every claim carries its source page.