Doubao Work was updated on November 12, adding support for multiple third-party large models and launching a limited-time special offer program. In addition to the two in-house models Doubao 2.1 Lite and Pro, users can now choose from models such as Kimi K3.1 Pro, GLM 5.3, and DeepSeek V4.1 Flash, expanding their options from the in-house lineup to a mainstream external lineup.

Because model usage costs differ, Doubao Work displays the quota consumption multiplier for each model on the model selection page. The lower the multiplier, the less quota is consumed. Users can make their own trade-offs based on task complexity, response speed, and cost-effectiveness. This approach turns differences between models into a quantifiable cost signal, making the trade-off between task results and resource consumption more transparent.

QQ20261112-140110.jpg

On the discount side, while simple, small tasks remain free, Doubao Work has introduced a limited-time promotion: during designated discount periods on weekdays and holidays, using participating models provides discounted quota consumption. The platform has also lowered the quota consumption multipliers for Doubao 2.1 Lite and DeepSeek V4.1 Flash, so the same tasks consume less and quotas last longer.

Doubao Work is an AI Agent office product launched by Doubao for productivity scenarios. It can execute tasks around user goals and deliver work results. Currently, new users who log in to the Doubao Work desktop version can receive a 30-day subscription benefit for free.

At a time when Agent products generally use a single in-house model as their foundation, opening third-party access and disclosing billing multipliers means the focus of competition is shifting from a contest of model capabilities to task scheduling flexibility and unit-cost efficiency. For users, it becomes possible to choose models by stage within the same workflow; for vendors, returning model choice and pricing transparency to users also continues the trend of AI office tools moving from a closed experience toward an open model marketplace.