GitHub Copilot’s new model carries a 73% lower list price than the model it follows. And annual subscribers get billed at a 0.25× premium request multiplier. The model is MAI-Code-1.1-Flash, Microsoft’s latest small-tier coding model. And GitHub’s changelog says it started rolling out across Copilot on August 11, 2026. Microsoft’s pitch is blunt: “better, faster, at a quarter of the cost.” What follows is what actually shipped, what it does to your bill. And the one default setting that will quietly cost Business and Enterprise teams the savings.
What MAI-Code-1.1-Flash Actually Is
This is the second iteration of Microsoft’s homegrown small coding model.
And the speed of the cycle is the real story.
MAI-Code-1-Flash landed in June 2026, rolling out to Copilot starting with VS Code, to a limited set of users, expanding gradually.
Microsoft introduced it as a model “built for fast, efficient assistance in everyday developer workflows.” By August 11, its successor was in production across Copilot.
The upgrade list is specific.
GitHub says 1.1-Flash adds native vision support for image understanding. And delivers improvements across coding quality, instruction following, tool use, and performance.
Microsoft adds that it produces higher quality code at 25% greater token efficiency than the June model.
The strategic tell came earlier, in GitHub’s Build 2026 recap, which called the Flash line “the first in a new wave of purpose-built coding models from Microsoft designed and tuned specifically for GitHub Copilot.” Read that again. The default AI coding assistant for millions of developers is moving off borrowed general-purpose models onto hardware-to-picker models Microsoft owns end to end. When the vendor controls the model, the serving stack, and the meter, pricing gets aggressive fast. That’s exactly what happened here.
The Pricing Math That Actually Matters
Two numbers do the work. GitHub says advances in model and serving efficiency enabled a 73% lower list price than MAI-Code-1-Flash. And for annual GitHub Copilot subscribers, the model is charged at a 0.25× premium request multiplier.
If you’ve never thought about multipliers, this is the week to start.
A 0.25× multiplier means one request against this model consumes a quarter of a premium request from your allowance. The expensive frontier models ration your attention because every prompt costs real budget. A model priced this low you just use, for the boring 80% of the job: autocomplete, boilerplate, renaming, CLI chores, first-pass scaffolding.
The savings compound. Microsoft’s claim stacks a quarter of the cost on top of 25% greater token efficiency, which means each request both costs less and burns fewer tokens doing the same work. I run a one-person agency and I think about model choice as a line item, not a preference. A cheap model I can spam without watching a meter beats a brilliant model I ration, for most of what client work actually consists of.
One caveat the changelog is quiet about: the 0.25× multiplier is stated for annual subscribers. If you’re on a monthly plan, check your own meter before you assume the same rate applies.
Where You Can Pick It, And Where You Can’t
Distribution is the other headline.
GitHub says you can select the model in the model picker across:
– Copilot CLI
– Copilot cloud agent
– GitHub Copilot app
– Copilot Chat on GitHub
– Visual Studio Code
– Visual Studio
– GitHub Mobile
– JetBrains IDEs
– Eclipse
– Xcode
That’s the IDE, the terminal, the phone, and the cloud agent. Compare that to June, when the first Flash model shipped to VS Code only, gradually, to a limited user set. Two months of iteration turned a single-surface experiment into a fleet-wide default candidate.
Access splits by tier.
Copilot Free and Student users get MAI-Code-1.1-Flash through auto model selection only.
So you take what the picker hands you.
Copilot Pro, Pro+, Max, Business, and Enterprise users can manually select it in addition to auto mode.
Now the catch from my title.
For Copilot Enterprise and Copilot Business plans, administrators must enable the MAI-Code-1.1-Flash policy in Copilot settings.
The policy is off by default. Which means the cheapest model in the lineup is invisible to most org seats until someone with admin rights flips a switch that nobody announced in a meeting.
Microsoft shipped a discount and defaulted most companies out of it.
The Read: Bundled Doesn’t Mean Best, And It Doesn’t Need To
Here’s my contrarian take.
Neither announcement ships a benchmark table.
Microsoft says “higher quality code” and GitHub says “improvements across coding quality” without head-to-head numbers.
And you should treat unbenchmarked vendor superlatives as marketing until you’ve tested them on your own repos. There’s also a fair fight happening in public between this model and cheap open-weight rivals. And I’m not convinced a bundled model wins on raw quality per dollar.
It doesn’t have to.
This model’s advantage isn’t the leaderboard, it’s placement: it sits in the model picker you already have open, priced low enough that switching costs nothing and staying costs almost nothing.
Incumbent distribution beats independent excellence more often than we’d like to admit, and this is a textbook case. Your team doesn’t evaluate models in a vacuum.
It uses whatever the picker defaults to.
And auto model selection is now steering Free and Student users straight at this one.
For small businesses and solo operators, the action is straightforward.
Individuals: open your model picker this week and pin MAI-Code-1.1-Flash for the low-stakes work, keep your premium budget for the hard problems where frontier models earn their multiplier.
Admins: enable the policy before your team assumes it’s already on, given that off-by-default is where savings go to die.
Everyone: run it against a real task from your own backlog before you trust the “better” half of Microsoft’s slogan.
The Verdict
MAI-Code-1.1-Flash shipped August 11, 2026, across every Copilot surface, at 73% off its predecessor’s list price with a 0.25× multiplier for annual subscribers. The economics are real, the quality claims are unproven.
And Business and Enterprise orgs get nothing until an admin acts.
GitHub is collecting reactions in the Community Discussion, which is worth a skim before you commit.
Do two things today.
Check your model picker and test this model on one tedious task you’d normally rush. If you run a Business or Enterprise org, flip the MAI-Code-1.1-Flash policy on now, as a discount your team can’t see is a discount you’re not getting. Then watch your next premium request statement, since that’s the only benchmark that pays your invoice.
