Doubao Work users have a new decision to make before handing over a document or spreadsheet task: which model to use, and how quickly it spends credits. The November 12 update adds Kimi K3.1 Pro, GLM 5.3 and DeepSeek V4.1 Flash among its third-party options, alongside a plan to label credit-consumption multipliers to help users choose.
Start with the desktop client, then check your identity
Mysteel's reproduction of Doubao's announcement says the additions sit alongside Doubao 2.1 Lite and Pro. Doubao says the model-selection page will label consumption multipliers, but the announcement does not confirm those labels are live. Lower coefficients mean fewer credits consumed. That is a spending indicator, not a quality ranking or a fixed quote for every job.
The standalone PC client is one place to begin. The announcement says new users can claim 30 days of subscription benefits by logging into Doubao Work on desktop; that does not mean unlimited free access to every model. Feishu's official product guide also lists access through Doubao and Feishu. It warns that a personal Doubao identity and a Feishu enterprise identity do not automatically share data, benefits or permissions. If an expected capability is missing, check the account, entry point and administrator settings before paying for another subscription.
A reduced multiplier is not unlimited access
The announcement separates two offers. Temporary multiplier reductions for Doubao 2.1 Lite and DeepSeek V4.1 Flash are already in place. Discounts during designated weekday and holiday periods are described as forthcoming. Do not assume both arrangements are active now, or that free small tasks mean unrestricted use of every model.
The announcement text reviewed does not provide a complete multiplier table, specific promotional hours, an end date or subscription prices in yuan. It therefore cannot establish the cash cost of producing a report, or the savings from a long subscription. Check the conditions displayed inside the product and the benefits attached to your account before starting work.
ToolAI's recommendation is to compare finished output rather than the lowest coefficient. Use the same non-sensitive material and delivery requirements for a small trial across available models. Record the actual deduction and the corrections each result needs. A cheap attempt that repeatedly needs reworking may not be the cheaper route to a usable file.
If your existing model already delivers reliably, there is no reason to rerun everything just because the menu has expanded. Change the default only after the new option demonstrates a useful difference in output or actual credit cost.