MAI-Code-1.1-Flash Is in GitHub Copilot — Act Before September 10

Microsoft pushed MAI-Code-1.1-Flash into GitHub Copilot on August 11, 2026, with a 73% price cut, native vision support, and benchmark gains, and will retire MAI-Code-1-Flash from all Copilot surfaces on September 10, 2026. The upgrade requires different actions by Copilot tier: Free and Student users need no action, Pro/Pro+/Max users can manually select the new model, and Business/Enterprise admins must explicitly enable the policy. Microsoft shipped the update under competitive pressure from DeepSeek V4 Pro and GLM-5.3, and early community feedback includes one developer reporting the model "misses code paths, misreads problems all the time and misdiagnoses frequently.

Microsoft pushed MAI-Code-1.1-Flash into GitHub Copilot on August 11, and if you’re still running MAI-Code-1-Flash, the clock is ticking: it gets retired from all Copilot surfaces on September 10 . The upgrade brings three things worth your attention — a 73% price cut, native vision support, and real benchmark gains on CLI and .NET workflows. The upgrade path differs by Copilot tier, so here’s exactly what you need to do. What Changed in MAI-Code-1.1-Flash MAI-Code-1.1-Flash is faster, cheaper, and can now process images. Tokens stream 25% faster and the model uses 25% fewer tokens to complete the same task. On Microsoft’s benchmarks https://microsoft.ai/news/mai-code-1-1-flash-br-better-faster-at-a-quarter-of-the-cost/ , Terminal-Bench 2.1 improved 22% in Copilot CLI and .NET tasks jumped 15%. Microsoft also reports a 4% rise in code survival rates and a 9% increase in developer return visits — both real-world signals that go beyond benchmark theater. The cost reduction is the headline: 73% cheaper than MAI-Code-1-Flash . Annual subscribers pay at a 0.25x premium request multiplier. Worth noting: Microsoft shipped this under competitive pressure. DeepSeek V4 Pro and GLM-5.3 have been eroding Copilot’s cost story, and 1.1-Flash is the direct counter-move. That’s good for developers regardless of Microsoft’s motives. Vision Support: What It Actually Means for Your Workflow MAI-Code-1.1-Flash is the first MAI-Code model to accept image inputs. Claude, GPT-4o, and Gemini have had this for a while — Microsoft was behind. Now it’s caught up. The practical use cases are front-end heavy: paste a screenshot and get HTML/CSS back, drop in an error dialog image for diagnosis, feed it an architecture diagram alongside your code, or use a whiteboard photo to kick off implementation. If you work on UI-heavy projects, this is useful. If you’re building backend services all day, you’ll probably ignore it. The September 10 Deadline: Who Needs to Do What MAI-Code-1-Flash goes dark on September 10, 2026 across all Copilot surfaces. Per the official deprecation notice https://github.blog/changelog/2026-08-11-upcoming-deprecation-of-mai-code-1-flash/ , what you need to do depends on your plan: Free and Student users: Nothing. You’ll be switched to automatic model selection automatically. Pro, Pro+, and Max users: You can manually select MAI-Code-1.1-Flash in the model picker now. After September 10, 1-Flash disappears from the list. Business and Enterprise: Admins must explicitly enable the MAI-Code-1.1-Flash policy in Copilot settings. It’s disabled by default. Once enabled, developers will see it in the Copilot Chat model selector in VS Code and on github.com. If you have workflows or CI integrations with a pinned model ID pointing to MAI-Code-1-Flash, update them before September 10. After that date, the old model becomes unavailable. No manual cleanup is needed — but if Enterprise admins don’t enable 1.1-Flash before the cutover, their teams will lose access to MAI-Code entirely until the policy is turned on. Early Community Feedback The model is four days old, so real-world feedback is thin. In the community discussion thread https://github.com/orgs/community/discussions/203965 , one developer noted the model “misses code paths, misreads problems all the time and misdiagnoses frequently.” That’s one data point, but it’s consistent with what you’d expect from a model that’s been in production for less than three months total. Microsoft is explicitly asking developers to report issues via the MAI-Code GitHub repo https://github.com/microsoft/MAI-Code . The rapid iteration cadence — 1-Flash to 1.1-Flash in about 10 weeks — is a good sign. Microsoft’s Superintelligence team says it’s building on developer feedback rather than benchmark optimization alone. The CLI and .NET gains exist because developers flagged those areas as weak spots. That kind of feedback loop is worth trusting, even if early versions still have rough edges. Bottom Line If you’re on Business or Enterprise, enable the 1.1-Flash policy this week — not September 9. If you’re on a paid individual plan, switch and test it alongside your current model for a few days. The cost and speed math is clearly better. The reliability story is still being written, but it’s heading in the right direction. MAI-Code-1.1-Flash is available across Copilot CLI, VS Code, Visual Studio, JetBrains IDEs, Eclipse, Xcode, GitHub Mobile, and the Copilot app. Full rollout details are in the GitHub Changelog https://github.blog/changelog/2026-08-11-mai-code-1-1-flash-available-in-github-copilot/ .