TLDR
- Microsoft unveiled MAI Code 2, a coding model it says nearly matches rivals at a fraction of the cost.
- A cybersecurity-specific model will go to a limited group of customers.
- New image-generation and transcription models also launched at Ignite.
- Mustafa Suleyman said domain-specific models keep costs low and named alignment his top priority.
- Meta’s Muse Wonder, Muse Spark 1.3 and Muse Ember are coming to Microsoft Foundry.
Microsoft (MSFT) rolled out a set of new AI models on Tuesday. The lineup covers coding, cybersecurity, images and transcription.
The company made the announcements at its Ignite conference in San Francisco. It is pushing to compete with Anthropic and OpenAI.
The headliner is MAI Code 2, a computer programming model. Microsoft says it almost matches rival models on performance at a fraction of the cost.
MICROSOFT ANNOUNCES 6 NEW AI MODELS AT IGNITE$MSFT unveiled six MAI models across coding, cybersecurity, image generation and voice on the same day its partner OpenAI released GPT-6.2.
1) MAI Code 2: Delivers 98% of GPT-6.1 Sol's performance at 40% of the cost. Now in GitHub… pic.twitter.com/yn6qA2K17c
— Wall St Engine (@wallstengine) November 17, 2026
Microsoft also showed a model built just for cybersecurity. It plans to offer it only to a limited group of customers.
The goal is to help with a surge in software vulnerabilities. AI tools are increasingly bringing those flaws to light.
A Domain-by-Domain Playbook
The launch also included new image-generation and transcription models. Microsoft did not stop at code and security.
Microsoft AI chief Mustafa Suleyman explained the thinking in an interview with Reuters. He said building technology for specific domains helps keep costs low.
He said that approach lets Microsoft match rivals and, in some cases, top model benchmarks. The company is not relying on a single model to do it.
“This isn’t just one model. It’s not a flash in the pan,” he said. He added that the approach is “a system, a hill-climbing machine” that balances performance and price.
In plain terms, Microsoft wants its AI to be both good and affordable. Suleyman framed that as a repeatable method rather than a one-time launch.
Safety Takes the Front Seat
Suleyman said the recent cybersecurity reckoning shaped his focus. He pointed to OpenAI agents hacking the Hugging Face platform as an example.
That episode pushed him to think about safeguards, not just performance. Speed alone was no longer the whole story for him.
His top priority is alignment. That is the research area focused on preventing AI from taking harmful actions and making sure it behaves as intended.
Ignite also brought news on outside models. CEO Satya Nadella said Meta’s latest AI models are coming to Microsoft Foundry.
Foundry is the company’s platform for building AI applications and agents on Azure. Nadella wrote on X, “We’re giving developers and enterprises even more choice and flexibility!”
Meta CEO Mark Zuckerberg said in a separate post that Muse Wonder, Muse Spark 1.3 and Muse Ember are “coming soon.” All three are headed to Foundry.







