Microsoft has unveiled new AI models for coding, cybersecurity and creative work, putting the cost of running artificial intelligence at the centre of its latest push for business customers.
Announced on November 17 at Ignite 2026, the new Microsoft AI models are designed for specific jobs rather than asking one large model to handle everything. The company’s pitch is straightforward: businesses should be able to automate more work without seeing their computing bills rise at the same pace.
The announcement adds another competitive challenge for OpenAI and Anthropic as companies compare AI tools on price, reliability and everyday usefulness.
Coding tools built for everyday work
Among the main releases are MAI Code 2 and Code 2 Flash. According to Microsoft’s Ignite announcement, Code 2 handles tasks such as finding software bugs, making changes across several files and understanding how different parts of a software project fit together.
Code 2 Flash focuses on faster, more frequent jobs, including explaining code, making targeted changes and fixing routine problems. Microsoft says both models are available through GitHub Copilot.
For a business, that distinction could matter. A simple software edit may not need the same computing resources as a complicated investigation into why an application keeps failing. Choosing a model suited to each task could help control spending. The Official Microsoft Blog
Lower prices are the main selling point
In its model launch details, Microsoft says MAI Code 2 delivers 98% of GPT-6.1 Sol’s coding quality at about 40% of its price, based on the company’s comparisons.
These figures are Microsoft’s claims about selected tests. They do not establish that every customer will achieve the same results or savings. Actual costs will depend on the work being done, how often the tools are used and how much checking or correction their output requires.
Still, the pricing message shows where Microsoft wants to compete. Alongside stronger capabilities, it is selling the idea that companies can afford to use AI more frequently. The lineup also includes new image-generation and audio capabilities, extending that approach beyond software development. Microsoft AI
Cybersecurity gets a specialist model
Microsoft also introduced MAI Cyber 2 Flash, built to help security teams find software weaknesses, check whether they pose a real risk and recommend fixes.
Its Foundry product update says the model will be available through Codename MDASH. The emphasis is on helping teams move through the full process of finding, confirming and repairing vulnerabilities.
That is a more practical goal than simply generating a long list of possible problems. Security teams still need to determine which findings deserve attention and whether a proposed repair actually works. Microsoft Community Hub
What matters next
The broader announcement links these models to Microsoft’s existing business software. Its Ignite overview describes tools that can carry out recurring tasks, alongside controls for monitoring and securing AI activity.
The business opportunity is clear: turn AI from something employees occasionally consult into something companies use throughout the working day.
The unanswered question is whether customers will see enough dependable output to justify wider deployment. Lower prices can make experimentation easier, but lasting adoption will depend on useful work completed, errors avoided and time saved. Those results will matter more than launch-day comparisons as the new Microsoft AI models reach customers.