# MAI-Code-1.1-Flash Ships: MAI-Code-1-Flash Retires Sep 10

MAI-Code-1.1-Flash is in Copilot: native vision, 22% better Terminal-Bench, a quarter of the cost. MAI-Code-1-Flash deprecates Sep 10.

By Neomanex. Updated 2026-10-07.

MAI-Code-1.1-Flash is a pure upgrade — same small-tier slot in GitHub Copilot, better everywhere Microsoft measured, at a quarter of the list price — and it's already rolling out. The model it replaces, MAI-Code-1-Flash, deprecates across every Copilot experience on **Sep 10, 2026**. That date is a hard migration deadline, not a suggestion. (Benchmarks are Microsoft's own; we haven't independently tested the model.)

## What happened

Microsoft shipped MAI-Code-1.1-Flash into [GitHub Copilot](/directory/github-copilot) on Aug 11, 2026, slotting it into the small-tier position [MAI-Code-1-Flash](/models/mai-code-1-flash) currently holds. The headline numbers (Microsoft AI, 2026):

| Metric | Change |
|---|---|
| Terminal-Bench 2.1 (Copilot CLI) | +22% |
| .NET tasks | +15% |
| Tokens per completed task | −25% |
| Streaming speed | +25% |
| List price vs 1-Flash | −73% |
| Code survival | +4% |
| Return visits | +9% |

Two genuinely new capabilities: **native vision** for image understanding, and the token-efficiency gains that drive most of the cost story.

## Why it matters

For most Copilot users this is a no-brainer swap — faster, cheaper, better at the tasks it was built for. The retiring [MAI-Code-1-Flash](/models/mai-code-1-flash) sits in our directory as **conditional**: a small, cost-optimized model purpose-built for Copilot autocompletions and refactors, not a frontier model. 1.1-Flash doesn't change that positioning — it sharpens it.

Two caveats before you bank the savings. First, Microsoft announced the price cut in **percentage terms only** — no exact per-token rate is published yet, so don't pencil in a specific $/1M figure until the Copilot pricing docs confirm it (GitHub Changelog, 2026). Second, annual Copilot subscribers are charged a **0.25× premium request multiplier** for the new model — "a quarter of the cost" describes the list price, and your effective rate depends on your plan.

## What changes for you

- **On MAI-Code-1-Flash today?** Move your workflows and integrations to MAI-Code-1.1-Flash before Sep 10, 2026. There's no reason to wait for the cutoff.
- **Business and Enterprise admins:** the new model's policy is **off by default** — enable it in Copilot settings.
- **Pro, Pro+, and Max users:** select MAI-Code-1.1-Flash manually in the Copilot model picker. Free and Student plans get it automatically via auto model selection.
- **Pricing-sensitive teams:** hold budget forecasts until Microsoft publishes a per-token rate, and factor in the premium-request multiplier on annual plans.

## FAQ

**What is MAI-Code-1.1-Flash?** Microsoft's updated in-house coding model for GitHub Copilot — a small, cost-optimized model that adds native vision and stronger token efficiency over MAI-Code-1-Flash.

**Do I have to do anything before Sep 10?** Yes, if you're on MAI-Code-1-Flash. It deprecates across every Copilot experience that day. Most users just reselect the new model; Business and Enterprise admins must enable the policy first.

**How much does it actually cost?** Microsoft has published only the "quarter of the cost" list-price claim, not a per-token rate. Wait for the Copilot pricing docs before forecasting.

Canonical: https://neomanex.com/news/mai-code-1-1-flash-launch