Skip to main content
Kaino.dev
Discover
Evals
News
Academics
Insights
Kaino.dev

Discover, evaluate, and compare AI tools, models, and agents.

Explore

  • Discover
  • Evaluations
  • News
  • Academics
  • Insights

Community

  • Twitter
  • YouTube
  • Instagram
Privacy PolicyTerms of Service

© 2026 Kaino.dev. All rights reserved.

Version 1.1.0
Microsoft’s MAI-Code-1-Flash Rolls Out in GitHub Copilot · News · Kaino
Microsoft’s MAI-Code-1-Flash Rolls Out in GitHub Copilot
Kaino
Jun 2Jun 2, 2026, 12:00 AM3 views

Microsoft’s MAI-Code-1-Flash Rolls Out in GitHub Copilot

Microsoft AI has introduced MAI-Code-1-Flash, a coding model now rolling out in GitHub Copilot. Microsoft says the model outperformed Claude Haiku 4.5 on tested coding benchmarks, including a 51.2% score on SWE-Bench Pro versus 35.2%, while using up to 60% fewer tokens on SWE-Bench Verified.

llmsintroducingcodeflashAILLMsGitHub CopilotMicrosoft AIcoding assistants

Microsoft AI has introduced MAI-Code-1-Flash, a coding-focused model that is now rolling out in GitHub Copilot.

A new coding model for Copilot

In a June 2 release, Microsoft AI said MAI-Code-1-Flash is being made available through GitHub Copilot. A separate GitHub Changelog post said the rollout is starting with VS Code and will expand to Copilot Free, Pro, Pro+, and Max plans over the coming weeks.

The company is positioning the model as a coding assistant option inside Copilot rather than as a standalone consumer product. GitHub’s changelog frames the launch as a Copilot availability update, while Microsoft AI’s announcement focuses on benchmark performance and efficiency claims.

Benchmark claims

Microsoft AI said MAI-Code-1-Flash outperformed Claude Haiku 4.5 across the coding benchmarks it tested. The company highlighted SWE-Bench Pro, where it said MAI-Code-1-Flash scored 51.2% compared with 35.2% for Claude Haiku 4.5.

Microsoft AI also said the model used up to 60% fewer tokens on SWE-Bench Verified. Token usage matters for coding assistants because lower token consumption can affect latency, context use, and operating costs, although Microsoft’s provided excerpt does not include deployment-level cost or speed figures for Copilot users.

The benchmark comparisons come from Microsoft AI’s own release. A Microsoft AI Build 2026 keynote transcript also referenced the MAI-Code-1-Flash benchmark claims and directed readers to the dedicated MAI-Code-1-Flash blog post for details.

Availability in GitHub Copilot

According to GitHub’s June 2 changelog, MAI-Code-1-Flash is rolling out first in VS Code. GitHub said it will become available to Copilot Free, Pro, Pro+, and Max plans over the coming weeks.

That staged rollout means users may not see the model immediately, depending on their Copilot plan and environment. The sources provided do not specify whether availability will vary by region, enterprise policy, or editor beyond the initial VS Code rollout.

What is known — and what is not

The public claims available in the provided sources are focused on coding benchmarks and Copilot availability. Microsoft AI’s announcement names Claude Haiku 4.5 as the comparison model in tested coding benchmarks and provides specific numbers for SWE-Bench Pro. It also states that MAI-Code-1-Flash can use up to 60% fewer tokens on SWE-Bench Verified.

However, the excerpts do not provide a full benchmark methodology, the complete list of tests, independent third-party validation, or details about how the model performs across different programming languages and real-world codebases. They also do not state whether MAI-Code-1-Flash will become the default Copilot model or remain an optional model choice.

For developers, the practical significance of the launch will depend on how the model performs inside everyday Copilot workflows: writing code, modifying existing projects, debugging, and following repository-specific context. GitHub’s changelog confirms the rollout path, but broader user experience details will become clearer as access expands across Copilot plans.

Why it matters

MAI-Code-1-Flash is another sign that Microsoft is continuing to integrate its own AI models into developer tools alongside models from other providers. Because GitHub Copilot is widely used by developers, even a staged model rollout can quickly put a new coding model in front of a large technical audience.

For now, the most concrete facts are the availability timeline and Microsoft’s benchmark claims: MAI-Code-1-Flash is rolling out in GitHub Copilot starting with VS Code, will reach Free, Pro, Pro+, and Max plans over the coming weeks, scored 51.2% on SWE-Bench Pro in Microsoft’s comparison, and used up to 60% fewer tokens on SWE-Bench Verified according to Microsoft AI.

Key takeaways
  • 1

    Microsoft AI has introduced MAI Code 1 Flash, a coding focused model that is now rolling out in GitHub Copilot.

  • 2

    A new coding model for Copilot In a June 2 release, Microsoft AI said MAI Code 1 Flash is being made available through GitHub Copilot.

  • 3

    A separate GitHub Changelog post said the rollout is starting with VS Code and will expand to Copilot Free, Pro, Pro+, and Max plans over the coming weeks.

Continue reading

Latest from Kaino News

Story pulse

Freshness

Jun 2

Views

3

Reading

3 min

Byline

Kainotomic Team

Utilities

Topics

llmsintroducingcodeflashAILLMsGitHub CopilotMicrosoft AIcoding assistants

Sources

Reference material and original reporting used in this story.

Microsoft AI

Published Jun 2, 2026, 12:00 AM

View source