Microsoft Brings MAI-Code-1-Flash to GitHub Copilot Business and Enterprise

Microsoft has made its MAI-Code-1-Flash coding model generally available for GitHub Copilot Business and GitHub Copilot Enterprise customers, as part of a broader recent expansion of its MAI AI model lineup. The update gives organizations access to Microsoft’s lightweight, first-party coding model inside managed Copilot plans, offering a faster and more cost-efficient option for everyday development tasks.

The rollout closes one of the remaining gaps in the model’s availability. While MAI-Code-1-Flash had already reached most GitHub Copilot experiences, including the CLI, cloud agent, GitHub.com chat, GitHub Mobile, Visual Studio, JetBrains IDEs, Eclipse, and Xcode, managed Business and Enterprise customers had been waiting for support.

Admins must enable the model

Organizations will not see the model automatically enabled. Business and Enterprise administrators must first turn on the MAI-Code-1-Flash policy in GitHub Copilot settings before developers can begin using it.

Once enabled, developers can select the model for supported Copilot experiences as part of their organization’s managed deployment.

Built for speed and efficiency

Microsoft introduced MAI-Code-1-Flash on June 2 as part of its internally developed MAI family of AI models. Rather than competing with the largest reasoning models, MAI-Code-1-Flash focuses on delivering quick responses for common coding workflows.

According to GitHub, the model is well suited for fast code completions, general-purpose coding tasks, code explanations, writing repetitive code, and high-volume developer workflows.

The emphasis is on responsiveness and lower operating costs instead of maximum reasoning capability.

Benchmark results

Microsoft says MAI-Code-1-Flash achieved a score of 51.2% on SWE-Bench Pro during its internal testing.

The company also claims the model required up to 60% fewer tokens on SWE-Bench Verified compared to other approaches, potentially lowering inference costs for organizations using GitHub Copilot under usage-based billing.

These benchmark figures come from Microsoft rather than an independent evaluation, so they should be viewed as vendor-reported results.

Usage-based pricing

GitHub bills MAI-Code-1-Flash through its usage-based pricing system. The published pricing is:

• Input tokens: $0.75 per million
• Cached input tokens: $0.075 per million
• Output tokens: $4.50 per million

As GitHub Copilot continues moving toward usage-based billing, smaller models like MAI-Code-1-Flash could help organizations reduce costs for repetitive coding and AI agent workloads.

Not designed for every development task

Although the model is optimized for speed, Microsoft does not position it as the best option for every scenario.

Large repositories, complex multi-file changes, advanced debugging, and security-sensitive code reviews may still benefit from larger AI models with stronger reasoning capabilities and broader context windows.

For many day-to-day development tasks, however, MAI-Code-1-Flash offers a balance between performance, speed, and efficiency that could make it an attractive default choice for enterprise development teams.

Microsoft’s expansion of MAI-Code-1-Flash to GitHub Copilot Business and Enterprise also comes as competition in the AI coding assistant market continues to intensify. In related news, OpenAI recently announced its new GPT-5.6 model family, introducing another generation of AI models aimed at coding, reasoning, and enterprise workloads.