Doubao Work integrates third-party models such as Kimi K3.1 Pro, GLM 5.3, and DeepSeek V4.1 Flash, clearly labels quota multipliers, and offers time-limited discounts to reduce enterprise AI office costs.Author and source: GeekPark
Doubao Work is now model-free.
On November 12, Doubao Work announced integration with multiple third-party models, including Kimi K3.1 Pro, GLM 5.3, and DeepSeek V4.1 Flash, enabling users to select the most suitable model for their tasks. The model selection page now displays the consumption multiplier for each model—the lower the multiplier, the less resource usage—helping users make informed decisions based on task requirements and cost considerations.
The platform is also offering two limited-time promotions: first, during designated hours on weekdays and holidays, using the promotional models grants a discount on quota consumption; second, the quota consumption multipliers for DouBao 2.1 Lite and DeepSeek V4.1 Flash are temporarily reduced, allowing you to complete the same tasks with less quota.
I. Complete the multi-model selection to solve the issue of "having options"
In this update, Doubao Work has integrated three third-party models—Kimi 3.1 Pro, GLM 5.3, and DeepSeek V4.1 Flash—in a single step, further expanding its model capability matrix.
Specifically, the three models each have distinct strengths and are suited to different workplace scenarios within an enterprise: Kimi 3.1 Pro excels in processing ultra-long contexts of up to millions of characters; GLM 5.3 focuses on Chinese language understanding and generation; and DeepSeek V4.1 Flash is designed for high response speed and cost efficiency, making it ideal for lightweight, everyday tasks.
Including the previously released third-party models and ByteDance’s proprietary models, such as Doubao 2.1 and Doubao 2.1 Lite, Doubao Work now offers a model portfolio covering three tiers: high-performance, balanced general-purpose, and lightweight efficiency.
Models with varying capabilities and pricing complement each other, broadly covering enterprises' AI needs across scenarios—from complex core tasks to everyday basic work.
In fact, DouBao Work had already opened up third-party model capabilities to enterprise customers. This round of expansion is not merely a quantitative increase, but a scenario-based enhancement grounded in real enterprise workflows.
Image source: DouBao Work official website
In addition, we noticed that Doubao Work has also updated its logo today, officially certifying the familiar Doubao character for professional office use.
Two, make the leverage multiplier transparent and return the cost decision-making power to users.
If model expansion addresses the question of "whether there are options," then transparently labeling the quota consumption multiplier addresses the question of "whether you understand your choices."
In today’s enterprise AI market, opaque pricing is a common challenge for many businesses. Billing rules vary across different models—some charge based on input and output tokens, others by the number of API calls—and the conversions between units are complex, making it difficult for average users to quickly determine the actual cost of using a model.
Many teams find it difficult to estimate the cost of individual tasks until they receive their bill and realize they’ve exceeded their budget, and it’s also hard to trace the specific components of those costs.
To address this issue, Doubao has uniformly labeled the quota consumption multipliers for each model on the model selection page—lower multipliers mean less platform quota is consumed for the same task. All models now follow a unified quota system, allowing users to clearly compare cost differences when choosing a model, enabling quick decisions based on task priority and budget, and facilitating departmental cost control for enterprises.
On top of the free basic small tasks, DouBao Work is simultaneously launching two limited-time promotional offers to further reduce usage costs:
First, time-specific discounts: During designated hours on weekdays and holidays, using the promotional model grants quota consumption reductions, balancing user costs with platform computational efficiency. Second, reduced multipliers for high-frequency models: For the two commonly used lightweight models—Doubao 2.1 Lite and DeepSeek V4.1 Flash—the consumption multipliers will be temporarily lowered, enabling the same tasks to consume less quota and enhancing the durability of quotas for all users.
As the foundational capabilities of large models gradually converge, the key to successful implementation lies in helping businesses adopt AI effectively and cost-efficiently. From offering a single model to providing a multi-model matrix, and from opaque pricing to transparent cost management, DouBao’s product evolution aligns with the trend of enterprise AI transitioning from “trial use” to “routine production tool.”
Looking one step further, when there are enough models and the multipliers are clear enough, users may no longer need to worry about choosing which model to use—platforms will automatically route and intelligently schedule tasks based on the type of task, becoming the next natural product direction.