The full story · 2 min
Microsoft has long relied on OpenAI for its most advanced AI models. At its Ignite conference, its own AI team, Microsoft AI, showed a set of cheaper models built for particular jobs.
The models
MAI Code 2 is for everyday coding at companies, and MAI Code 2 Flash is a smaller model for large amounts of routine coding. MAI Cyber 2 Flash is for security agents that watch systems all the time. MAI Image 2.7 makes and edits pictures, and MAI Realtime 1 is a live voice model that people can try at MAI Labs.
Microsoft also updated Decision, its model that picks the right model for a task, to version 1.1. It says companies will be able to license its model weights and adapt them with their own data, inside Microsoft’s Azure cloud.
From Microsoft AI’s post and GitHub’s price list of November 17.
Microsoft’s claims
Microsoft says MAI Code 2 reaches 98% of the coding quality of OpenAI’s GPT-6.1 Sol at about 40% of the price. The figure is an average of four coding tests that Microsoft chose. Microsoft also says it reaches 96% of Anthropic’s Claude Opus 5.5 at about 5% of the price.
For the other models, Microsoft says MAI Code 2 Flash matches GPT-6 Luna at 65% of the price. It says MAI Cyber 2 Flash beats GPT-6 Astra on a security test at less than 2% of the cost per task. It also says MAI Image 2.7 ranks first for image editing on Artificial Analysis.
Microsoft’s comparisons rely on selected tests, and some of its price advantages come from promotional discounts, Axios notes.
Where, and at what price
MAI Code 2 is in GitHub Copilot’s paid plans, and MAI Code 2 Flash in all of them, including the free plan, GitHub says. MAI Code 2 Flash is also in a limited preview on Microsoft Foundry, Microsoft’s platform for developers.
In Copilot, MAI Code 2 costs $0.50 per million input tokens and $2.50 per million output tokens, against $2 and $10 for GPT-6.1 Sol. These are promotional prices until May 17, 2027. After that, MAI Code 2 costs $1 and $5.
“My team is building true superintelligence,” Suleyman told Axios. He gave no date for Microsoft’s own general-purpose model.
Sources
Every fact in this story comes from the sources below. Open them to check our work.
- 1Primary source · November 17, 2026Seven new world class models at scalable price-performance Microsoft AI
- 2
- 3Research · November 17, 2026Microsoft aims to challenge OpenAI, Anthropic with cheaper, specialized AI models Axios
- 4
- 5
- 6Research · November 17, 2026Microsoft’s AI pitch at Ignite: Cheaper models as business ‘token spend is going through the roof’ Yahoo Finance
AI desk: Developer tools and tests · published November 17, 2026, 23:56 CET. This story was generated by AI systems under the editorial rules of Silicon AI News and checked against the sources it links. No person read it before it was published. The site’s owner and editor: Y. D. Daniel, Budapest, Hungary. Found a mistake? Email editor@siliconainews.com. How we report
More: words to know, tags and how to cite this story
- GitHub Copilot
- Microsoft’s AI helper that writes and checks code inside programmers’ tools.
- Model weights
- The learned numbers inside an AI model. With them, a company can adapt and run it.
- Benchmark
- A standard test used to compare AI models.
Microsoft says its MAI Code 2 rivals GPT-6.1 Sol for less. Silicon AI News, November 17, 2026. https://siliconainews.com/stories/microsoft-mai-code-2-models