Doubao Work adds Kimi K3.1 Pro, GLM-5.3 and DeepSeek-V4.1-Flash for team and enterprise plans
ByteDance's Doubao Work said on October 29 that it is opening third-party models to business customers. Kimi K3.1 Pro, GLM-5.3 and DeepSeek-V4.1-Flash are the first batch, available on team and enterprise plans and billed against existing AI credits.
Doubao Work, ByteDance's AI office agent app, said on October 29 that it is opening third-party models to business customers. The first batch includes Kimi K3.1 Pro from Moonshot AI, GLM-5.3 from Zhipu and DeepSeek-V4.1-Flash. All three appear in the app's model picker next to ByteDance's own Doubao models, according to a report from Pandaily.
What changed
The update applies to Doubao Work's team and enterprise subscription plans and is switched on by default. Companies do not need to buy access from each model provider or file a separate application. Usage is deducted from the AI credits an organization already holds in Doubao Work.
Administrator controls
Administrators decide who can use which model. Access can be granted to:
- Individual members
- Departments
- Groups
- The whole company
Admins can also enable every model from one provider at once, or set up a single model on its own.
Models in the first batch
| Model | Provider |
|---|---|
| Kimi K3.1 Pro | Moonshot AI |
| GLM-5.3 | Zhipu |
| DeepSeek-V4.1-Flash | DeepSeek |
Context window sizes, per-model credit rates and benchmark scores for these models in Doubao Work have not yet been disclosed. ByteDance did not say which other third-party models would follow, describing the current lineup as the first group of models to join.
Context
Pandaily reports that the app was until now largely a showcase for ByteDance's Doubao models inside an agent that can work across files, browsers and desktop apps. Pandaily describes the move as one that makes the agent layer, rather than the underlying model, the main product. It also notes that teams can now choose a model by task, for example a coding-focused model for engineering work and a different model for long documents.
Pandaily places the change within a wider pattern among Chinese desktop agent products. Xiaomi's MiMo Desktop added DeepSeek, GLM, Kimi and Qwen models this week, and several other agent tools have made similar moves.
Doubao Work has been shipping features at a steady pace. Recent additions include:
- Saving recorded computer actions as reusable skills
- One-click import of chat history, skills and memory from other AI tools
- An infinite creation canvas
- Parallel agents
- Local GUI control on Mac
What this means
For organizations already on Doubao Work's team or enterprise plans, trying Kimi K3.1 Pro, GLM-5.3 or DeepSeek-V4.1-Flash requires no new vendor contracts or applications. Usage draws on the same AI credit balance. Because access is on by default and administrators can scope it by member, department, group or company, IT teams can restrict specific models without blocking the rest.
The change also shifts what Doubao Work sells. ByteDance's own Doubao models now sit in the same picker as models from Moonshot AI, Zhipu and DeepSeek, so the product's value rests on the agent features (skills, parallel agents, desktop control, memory import) as much as on the model behind them. Whether the third-party models carry different credit costs than Doubao models is not yet disclosed.
Related Articles
Tencent launches Buddy One, linking Docs, Meeting, Drive and ima under one WorkBuddy agent
Tencent launched Buddy One at its 2026 Global Digital Ecosystem Summit on October 29, connecting Tencent Docs, Tencent Meeting, Tencent Drive and ima to a single WorkBuddy agent with one account and shared AI credits. It also released Tencent Drive, cloud storage for agents that connects to third-party tools through a CLI and the Model Context Protocol.
ByteDance's TRAE launches TRAE Bao, a personal assistant with cross-project memory
ByteDance's TRAE launched TRAE Bao on October 29, a personal assistant that keeps memory across a developer's projects and tasks. It pairs a global assistant with a per-project assistant and can take tasks from Feishu and WeChat. The current version supports local tasks only.
GitHub Copilot in VS Code adds more choice in models, agent locations and workflows in Universe releases
GitHub published its Universe releases for GitHub Copilot in Visual Studio Code on 2026-10-28. The post says this year's focus was giving developers more choice in which models they use and where their agents run.
Anthropic makes Claude Enterprise smart reports, pooled budgets and in-app surveys generally available
Anthropic has made smart reports, pooled budgets and in-app surveys generally available for Claude Enterprise plans. All three launched in beta in September and are aimed at tracking AI usage, cost and employee feedback, according to ZDNET.
Comments
Loading...