Microsoft's new business AI runs on Alibaba's Chinese model. Its own comes later.
Microsoft released Microsoft-Decision-1, a model that returns structured answers software can act on instead of open-ended text, The Register reported on 10 October.
The first version is built on Qwen3.5-9B, an open-weight model from Alibaba Cloud. Microsoft says it will "soon rebase" Decision-1 on its own models and those from OpenAI.
Microsoft prices input at $0.042 per million tokens and charges nothing for output. On its own benchmarks it reports 83.5% accuracy across 36 tests and a cost more than 20 times below OpenAI's GPT-6 Sol for text classification. It is available on Microsoft Foundry and is coming to OpenRouter.
Berlin search engine Ecosia moved from Mistral to Chinese open models including Qwen this week; its CEO Christian Kroll said running costs halved, heise reported.
A team that deploys it to route invoices or tickets would run a Chinese base model in that step until the rebase arrives. Microsoft has given no date for it.
Would your compliance team sign off on a Qwen-based model in an approval workflow while the switch has no date?
Sources
Our file on Microsoft
- 12 Oct
Microsoft's CEO told firms to treat AI as risky insiders. Keep an off switch.
- 11 Oct
Fujitsu gave 100,000 staff Copilot. AI now handles about 75% of help desk triage.
- 10 Oct
Microsoft 365 families will lose up to 4 TB. Microsoft calls it flexible.
- 10 Oct
Microsoft billed a startup $17,600 for Claude. Its $21,168 in credits did not apply.
- 9 Oct
Washington barred Microsoft, Infosys and six others from green-card sponsorship. Pending cases stopped too.
Every story here is open to read. The ERP LEADERS brief goes one step further.
The week in enterprise software, in one email: the stories that mattered, the failures with their figures, one case taken apart. Read a sample or sign up for the brief.
Welcome back. · Issue 01
