GitHub Copilot Adds MAI-Code-1.1-Flash as Microsoft Cuts Coding Model Prices by 73%

Microsoft's latest small-tier coding model, MAI-Code-1.1-Flash, began rolling out in GitHub Copilot on August 11, adding native vision support for image understanding plus improvements in coding quality, instruction following, tool use and performance, the company's changelog said.

Faster and Cheaper

Compared with MAI-Code-1-Flash, introduced at Microsoft Build in June, the new model scores 22 percent higher on Terminal-Bench 2.1 and improves .NET-related tasks by 15 percent, while token streaming speed is up 25 percent and tokens consumed per task are down 25 percent. Microsoft cut token prices across all tiers by about 73.3 percent to roughly one-quarter of the original model's — input at $0.20 per million tokens, with a 0.25x request multiplier for annual Copilot subscribers.

Developer Reach

MAI-Code-1.1-Flash is available through Copilot in VS Code, Visual Studio and JetBrains IDEs, giving developers a low-cost option alongside GPT-5.6 and Anthropic's Claude models in the assistant. The release is part of Microsoft's push to make Copilot the default agentic coding layer, with a unified Copilot super-app rollout across consumer and enterprise versions beginning in mid-August.