Artificial intelligence news.
Straight from the original announcements.

Microsoft offers businesses control over customized MAI models

November 17, 2026, 19:17By the AI News Bites Editorial TeamSource: Microsoft AI
Microsoft offers businesses control over customized MAI models
Image: Microsoft AI

Microsoft is introducing MAI Licensed Weights, a service that will let organizations deploy and customize Microsoft AI models inside their own Azure environments. The company says customers will retain control over access, deployment, upgrades and adaptations to their models.

Key points
  • MAI Licensed Weights will bring Microsoft AI models into customers’ Azure environments.
  • Organizations will be able to customize models and control access, deployment and upgrades.
  • AT&T reported MAI performance at 1/50th of the cost of GPT-6.1 Sol for standard 5G network tasks.
  • Woven by Toyota reported ready-to-merge code fixes in 44% of evaluated cases.
  • The offering is coming soon through Microsoft’s Frontier Company engagements.

Models deployed in customers’ Azure environments

MAI Licensed Weights is designed for businesses that want greater control over the AI models embedded in their products and workflows. Microsoft says organizations will be able to adapt its models to their own business needs, while integrating proprietary knowledge and expertise in their Azure environments.

According to Microsoft, customers will own their data and adaptations. The company says it will use those materials to train its models only with the customer’s express agreement. Organizations will also be able to decide who can access their models and when they are upgraded, allowing them to manage changes according to their own operational requirements.

Customer examples show task-specific gains

Microsoft argues that the most useful model for a business is not necessarily the largest general-purpose system. Instead, it says customers can achieve stronger results on specific tasks by adapting MAI models to their proprietary workflows, while keeping deployment costs lower.

AT&T used customized MAI models to analyze 5G networks. Microsoft says MAI outperformed GPT-6.1 Sol on standard tasks at 1/50th of the cost. For more complex tasks, the company says the models reached frontier-level performance at a significantly lower cost.

Woven by Toyota used MAI models to fix software code. The company says the models produced ready-to-merge fixes in 44% of cases where its existing frontier models required human intervention, at less than one-tenth of the cost.

Alaska Airlines used MAI Code 2 Flash for aircraft maintenance report classification. According to Microsoft, tuning improved accuracy from 81.2% to 91.1%, outperforming leading frontier models at a fraction of the cost.

Availability will begin through Frontier Company engagements

Microsoft says MAI Licensed Weights is coming soon through its Frontier Company engagements. Organizations interested in the offering are directed to contact their Microsoft account team.

The company cautions that the reported results come from customer-specific evaluations of models adapted for particular tasks. Performance and cost advantages can vary depending on the workload, model configuration and evaluation method. Microsoft also says qualified employees retain responsibility for final decisions.

For businesses considering AI for specialized operations, the announcement focuses on a tradeoff between model control, task performance and operating cost. The customer examples suggest that adapting a model to internal expertise can produce results suited to a particular workflow, while the licensing approach is intended to let organizations manage their data, access policies and model upgrades within Azure.

Related news

AWS brings full fine-tuning to SageMaker AI’s serverless customization
Products and toolsNovember 19, 2026, 01:15

AWS brings full fine-tuning to SageMaker AI’s serverless customization

AWS has added Full Fine-Tuning (FFT) support to SageMaker AI Serverless Model Customization, allowing businesses to update all model parameters without provisioning GPUs…

Mastra launches Gateway for access to more than 500 AI models
Products and toolsNovember 19, 2026

Mastra launches Gateway for access to more than 500 AI models

Mastra has launched Mastra Gateway, a platform that gives developers access to more than 500 language, voice, video and decision models through a single API key and…

AWS Launches SageMaker JumpStart Leaderboard for Model Comparisons
Products and toolsNovember 18, 2026, 23:31

AWS Launches SageMaker JumpStart Leaderboard for Model Comparisons

AWS has launched the Amazon SageMaker JumpStart Model Leaderboard, a SageMaker Studio view that compares open-weight models on quality, performance and estimated cost.…