Qwen Office adds 10 large models including Kimi, GLM, DeepSeek
11/16/2026 — 11/16, 20:19·2 sources (all)·2 reports (all)
Key takeaways
- Alibaba's Qwen Office announced on November 16 the addition of 10 large models, including Qwen4-Max and third-party models from Kimi, GLM, D…
Background & analysis
On November 16, Alibaba's AI office product Qwen Office announced the addition of 10 large models, including the newly released flagship model Qwen4-Max alongside third-party models from providers such as Kimi, GLM, DeepSeek, MiniMax, and Step. With the update, Qwen Office formally opened model selection: users of both the personal and enterprise standard editions can now switch models based on task complexity, response speed, and Token cost. Previously, the platform mainly offered three tiers, namely Standard, Advanced, and Qwen3.8-Max.
According to the disclosures, Qwen4-Max is the first model in the Qwen4 series. It adopts a brand-new Qwen4 series architecture, with 2.4 trillion total parameters and 710 billion parameters activated per Token. It features native visual understanding and supports a 100万 Token context window. Alibaba's published benchmark results say Qwen4-Max improves noticeably over Qwen3.8-Max on multiple professional Agent benchmarks, and in the Work360 evaluation its average cost per task dropped from 26.58 CNY to 11.16 CNY.
The rollout of external models has been uneven. As early as August 14, Qwen Office had announced the integration of Zhipu's GLM-5.3 and DeepSeek V4 Pro, but only three days later, on the evening of August 17, both external models disappeared from the product, and the model selector returned to a setup based on capability tiers and Qwen3.8-Max. Alibaba never publicly explained the removal. Roughly three months later, Qwen Office brought back a batch of 10 models, a move that came just four days after ByteDance's Doubao Work fully opened third-party models to individual users. Reports note that Tencent WorkBuddy, ByteDance's Doubao Work, Baidu's Kuku AI, and Alibaba's Qwen Office have all moved toward a multi-model aggregation approach. At present, the 10 new models are live and model selection remains open, while Alibaba has not accounted for the earlier withdrawal of the two third-party models.
AI-generated from 2 reports · updated 3 hours ago
Latest turnAlibaba's Qwen Office has added 10 large models, including the newly released Qwen4-Max and third-party models from Kimi, GLM, DeepSeek, MiniMax and Step, opening up model selection. Users can switch models based on task complexity, response speed and token cost. Qwen4-Max uses a new Qwen4 architecture with 2.4T total parameters, 71B activated per token, and a 1M-token context window.
Related AI tools
- Heat index
- 118
- All sources
- 2
- All reports
- 2
- First seen
- 21 hours ago
Reports on this story headlines open the original
Alibaba's Qwen Office has added 10 large models, including the newly released Qwen4-Max and third-party models from Kimi, GLM, DeepSeek, MiniMax and Step, opening up model selection. Users can switch models based on task complexity, response speed and token cost. Qwen4-Max uses a new Qwen4 architecture with 2.4T total parameters, 71B activated per token, and a 1M-token context window.
智东西AI score 67Alibaba's Qwen Office added 10 large models, including its own Qwen4-Max plus third-party options from Kimi, GLM, DeepSeek, MiniMax and Step, letting individual and standard enterprise users choose models by task, speed and token cost. The move comes four days after ByteDance's Doubao opened third-party models to individual users.
华尔街见闻AI score 57
Other stories people are talking about
- Heat index 921Alibaba releases MoE model Qwen4-Max with 1M token context for low-cost, long-running tasks17 sources
- Heat index 80Tutorial: Estimating VRAM and Bandwidth for Running LLMs Locally1 sources
- Heat index 80Test: GLM and DeepSeek Run Up to 2.3x Faster on Mistral1 sources
- Heat index 144SurgeDeepSeek tightens free-tier limits on long chats and daily usage2 sources
- Heat index 50DeepSeek Harness Web version tutorial: dsh web and local Ollama1 sources
How is heat calculated?About the methodHide
Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.
This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.
- Surge
- Discussion rising fast
- New
- First report within 6 hours
- Rising
- Still gathering discussion

