Microsoft just cut the price of its coding AI by roughly three quarters and made it faster in the same breath.
The new model is called MAI-Code-1.1-Flash, and it rolled out across GitHub Copilot this week, in VS Code, Visual Studio, the JetBrains IDEs, the Copilot CLI, even GitHub Mobile. Input tokens now run 20 cents per million, cached input runs 2 cents, and output costs a dollar twenty. Set that next to where the original model started and you are looking at close to a 73 percent price cut.
It costs less and it tests better too. Microsoft says it scores 22 percent higher on a coding benchmark called Terminal-Bench when used through the Copilot CLI, and it runs 15 percent better specifically on dot NET work, which tracks given who built it. Tokens stream a quarter faster, and the model needs a quarter fewer tokens to finish the same task.
None of those numbers mean much sitting alone on a spec sheet. Put cheap and fast together though, and it changes how people actually build software. When a coding model gets this inexpensive, developers stop rationing it. You stop thinking twice before asking it to check one more thing, refactor one more function, run one more test pass. The friction that used to make you second guess whether a task was worth calling the AI for just gets smaller.
Microsoft has been open that part of this is a response to how fast and cheap Chinese coding models have gotten this year. Competition dragging prices down for the rest of us is not a bad trade at all. If this pace keeps up, having an AI pair programmer stops being a cost decision you have to think about, it just becomes the default.
——
Follow: @Ali Demi
Book your free AI clarity call, NOW!
https://buff.ly/TpWy277
——
Sources:
https://microsoft.ai/news/mai-code-1-1-flash-br-better-faster-at-a-quarter-of-the-cost/
https://github.blog/changelog/2026-08-11-mai-code-1-1-flash-available-in-github-copilot/
https://www.neowin.net/news/microsoft-releases-mai-code-11-flash-coding-model-to-better-compete-with-chinese-models/
Repost this. Thanks.

