GitHub Unveils Copilot Fusion to Cut AI Coding Costs

Microsoft-owned GitHub has introduced Copilot Fusion, a new AI-powered system designed to help developers reduce coding costs by automatically selecting and coordinating different AI models for software development tasks. Unveiled at the GitHub Universe event in San Francisco, the system routes tasks to suitable models and manages their collaboration instead of functioning as a standalone AI model. GitHub claims that Fusion Max scored 60 on the Artificial Analysis Coding Agent Index, compared with 58 for Claude Code powered by Opus 5.5 Max. The company also says it completed tasks 40% faster while using approximately half the API cost and tokens.

Developers can choose between three effort levels — Standard, High and Max — while Copilot Fusion manages model selection and workflows such as cascading and critique reviews. Microsoft CEO Satya Nadella stated that the system delivers frontier-level performance up to 40% faster and at up to 77% lower cost. GitHub also announced its Copilot app, which connects AI agents and tools across projects. Enterprise customers can access Copilot features without a separate subscription, paying only for the AI credits they consume. GitHub currently has more than 225 million developers worldwide, including 32 million in India, making it one of the largest platforms for software development.