Microsoft Brings MAI Code 1.1 Flash Local Inference to GitHub Copilot

Microsoft has announced the integration of MAI Code 1.1 Flash into GitHub Copilot, enabling local inference for coding workflows. This update allows developers to run AI-assisted coding tasks directly on their devices, reducing reliance on cloud-based models. The model features a 256,000-token context window and utilizes 137 billion parameters, optimized via quantization and speculative decoding. Users can leverage 'Auto orchestration' to dynamically route requests between local and cloud environments or manually select local execution via Windows ML or local endpoints. While the performance metrics on high-end hardware are impressive, Microsoft has yet to clarify the specific hardware requirements or commercial pricing models for local usage. The rollout is scheduled to begin with a limited audience by the end of October 2026, marking a significant step toward hybrid AI development environments that balance on-device privacy and performance with cloud-based capabilities.