Sign in
Pending benchmarksTracking since Nov 17, 2026

Is MAI Code 2 Flash state of the art?Pending benchmarks.Announced Nov 17, 2026; no independent results yet.

Microsoft AI's high-volume coding model, announced November 17, 2026, for code explanations, targeted edits, routine fixes, and rapid iteration. Available in GitHub Copilot for Free, Student, Pro, Pro+, Max, Business, and Enterprise subscribers on eligible annual plans. Official GitHub pricing is promotional through May 17, 2027 at $0.05 input, $0.005 cached input, and $0.25 output per 1M tokens, then $0.10 / $0.01 / $0.50. Official materials also describe a limited-access preview in Microsoft Foundry, which is not a public self-serve API. No open weights.

Announced facts

The details so far

Lab
Microsoft AI (MAI)
Announced
Nov 17, 2026
API
Not yet · Official model blog: limited-access preview in Microsoft Foundry; a later release on OpenRouter, Vercel, Hermes, Open Code, and Kilo is described as coming. No public self-serve Foundry API is stated
Open weights
Closed · Not released. MAI Licensed Weights is a partner Azure offering and does not count as ungated public weights
Focus
High-volume everyday coding: explanations, targeted edits, routine fixes, and rapid iteration (official model blog)
Availability
GitHub Copilot for Copilot Free, Student, Pro, Pro+, Max, Business, and Enterprise, on eligible annual plans, in the same Copilot surfaces listed on the official changelog
Price (GitHub Copilot)
Promotional official GitHub pricing through May 17, 2027: $0.05 input, $0.005 cached input, and $0.25 output per 1M tokens. Starting May 17, 2027: $0.10 / $0.01 / $0.50
Context
Not stated on the official announcement or Copilot pricing pages
Related
Separate from MAI Code 2 and from MAI-Code-1.1-Flash

Tags marked unsure are not stated on the lab's own pages.

Next steps

What would let us call it

Verdicts use only independent boards we already track. Two consecutive snapshots are required before any crown.

  • Publicly announced with a source
  • Usable by the public (API or weights)
  • Listed on an independent board we track
  • Enough results for a rank in a category
  • Holds across two snapshots (crowns only)
Self-reported

What the lab claims

Self-reported. Shown for context; not used in any verdict.

Lab claim: self-reported official blog comparison, across DeepSWE 1.1, Terminal Bench 2.1, and SWE-Atlas, that MAI Code 2 Flash offers comparable performance to GPT-6 Luna at 65% of the cost and to Claude Haiku 5.5 at 20% of the cost. Official footnotes say all models were run at max reasoning effort, and the cost comparison uses MAI Code 2 Flash's 50% promotional price.

Receipts

Sources

LABSeven new world class models at scalable price-performance (Microsoft AI, Ignite)microsoft.ai/news/microsoft-ignite-2026/ · Nov 17, 2026LABMAI Code 2 and Code 2 Flash (Microsoft AI)microsoft.ai/news/introducing-mai-code-2-and-mai-code-2-flash/ · Nov 17, 2026LABMAI Code 2 and MAI Code 2 Flash in GitHub Copilot (GitHub Changelog)github.blog/changelog/2026-11-17-mai-code-2-and-mai-code-2-flash-in-github-copilot/ · Nov 17, 2026LABModels and pricing for GitHub Copilot (GitHub Docs)docs.github.com/en/copilot/reference/copilot-billing/models-and-pricing · Nov 17, 2026LABMicrosoft AI on Xx.com/MicrosoftAI/status/2122762037927322044 · Nov 17, 2026
Editorial

Compare when it lands

Unscored: MAI Code 2. A scored comparison appears once both are on a shared board.

Tracking ends Feb 15, 2027 unless the API or weights become available.