Artificial intelligence news.
Straight from the original announcements.

Microsoft details MAI models focused on enterprise cost and control

November 17, 2026, 19:14By the AI News Bites Editorial TeamSource: Microsoft AI
Microsoft details MAI models focused on enterprise cost and control
Image: Microsoft AI

Microsoft has outlined a family of MAI models designed to give enterprises lower-cost performance for coding, cybersecurity, image editing and other workloads. The company also introduced MAI Licensed Weights, which lets customers run customized models in their own Azure environments.

Key points
  • MAI Code 2 is claimed to deliver 98% of GPT 6.1 Sol's performance at 40% of the cost.
  • MAI Cyber 2 Flash is claimed to beat GPT 6 Astra on Cyber Gym at less than 2% of the cost per task.
  • MAI Image 2.7 is presented as an image editing model priced at a fifth of the competition's cost.
  • MAI Transcribe is deployed to 10 million patient Dragon Copilot encounters a month.
  • MAI Licensed Weights lets customers customize and run models in their own Azure environments.

Models built around enterprise workloads

In the Microsoft Ignite 2026 MAI keynote transcript, Microsoft AI said its goal is to build models that keep developers and organizations at the frontier while addressing rising token costs. The company said it is prioritizing price performance alongside model quality, rather than optimizing only for the most demanding frontier capabilities.

For coding, Microsoft said MAI Code 2 delivers 98% of the performance of GPT 6.1 Sol across four widely used benchmarks for everyday enterprise coding tasks, at 40% of the cost. The company said MAI Code 2 costs 5% as much as Opus 5.5 on that comparison. Code 2 Flash, meanwhile, matches 6 Luna and Haiku 5.5 on these tasks at 65% and 20% of their prices, respectively.

Microsoft said these models are deliberately focused on routine customer coding work rather than fundamental science research or very long horizon agentic tasks. It said this approach avoids making customers pay for capabilities they may not need on every task.

Cybersecurity, multimodal AI and medicine

Microsoft also presented MAI Decision 1.1, which provides model routing inside applications. According to the company, it beats Jev, Google and OpenAI on the AA Decision Index.

In cybersecurity, Microsoft described MAI Cyber 2 Flash as its response to attacks and exploits becoming cheaper and easier to carry out. The company said the model is ahead of GPT 6 Astra on Cyber Gym end-to-end at less than 2% of the cost per task. Microsoft attributed the model's capabilities to its proprietary data and security expertise, saying it can detect vulnerabilities and recommend fixes.

The company said it is now the number two lab in image generation and described MAI Image 2.7 as the world's number one image editing model at a fifth of the competition's cost. Adobe, Sports Illustrated and Manus are cited as customers that say the model provides professional quality and precise, responsive edits.

Microsoft also said MAI Transcribe has been deployed to 10 million patient Dragon Copilot encounters a month while halving the error rate. It described its transcription model as the fastest and most accurate in the world.

Licensed weights put deployment under customer control

The keynote's central enterprise message was control over proprietary data, workflows and model adaptations. Microsoft said customers do not want to put their company's intellectual property into someone else's model, and announced MAI Licensed Weights as a way to address that concern.

With the offering, customers can take a frontier MAI model, customize it with proprietary data if they choose, and run it securely in their own Azure environment. Microsoft said customers control access, while their data and adaptations remain entirely theirs. The company will work with customers through Frontier Company to deliver the custom model and embed it in their own products and agents.

Microsoft framed the MAI portfolio as a set of specialized models intended to make AI deployment more scalable, while giving organizations control over the models supporting their most valuable data and workflows. Its stated promise is “Your IQ. Your Agents. Your Weights. All under your control.”

Related news

Microsoft AI unveils seven models targeting lower-cost AI workloadsTop story
ModelsNovember 17, 2026, 19:16

Microsoft AI unveils seven models targeting lower-cost AI workloads

Microsoft AI says it is releasing seven new models designed to combine high capability with lower price-performance for developers and enterprises. The lineup targets…

Microsoft unveils MAI Cyber 2 Flash for lower-cost security tasks
ModelsNovember 17, 2026, 19:15

Microsoft unveils MAI Cyber 2 Flash for lower-cost security tasks

Microsoft has introduced MAI Cyber 2 Flash, a specialist security model designed to find vulnerabilities, recommend fixes and investigate threats at lower cost. The…

Microsoft introduces MAI Code 2 models for lower-cost enterprise coding
ModelsNovember 17, 2026, 19:15

Microsoft introduces MAI Code 2 models for lower-cost enterprise coding

Microsoft has introduced MAI Code 2 and MAI Code 2 Flash, two coding models designed for everyday enterprise development tasks at lower costs than competing models. The…