GitHub releases Copilot Fusion, which automatically orchestrates multiple models to improve coding efficiency and cost performance. GitHub officially releases Copilot Fusion at the Universe2026 conference. This is an intelligent orchestration layer that can automatically route and combine OpenAI, Anthropic, Goo

GitHub releases Copilot Fusion, which automatically orchestrates multiple models to improve coding efficiency and cost performance. GitHub officially released Copilot Fusion at the Universe2026 conference. This is an intelligent orchestration layer that can automatically route and combine multiple AI models such as OpenAI, Anthropic, Google to complete coding tasks, without the need for developers to manually select models. This feature integrates existing automatic model selection technology with Project HydraFusion multi-model orchestration capabilities previewed in September. Developers only need to select Copilot Fusion in the model selector, and the system will automatically decide to use a single model, Cascade, or Critique execution modes based on the complexity of the task, and match the three effort levels of Standard (focusing on efficiency and cost), High (enhancing reasoning), or Max (pursuing the strongest capabilities) to avoid unnecessary overhead for simple tasks. In the Artificial Analysis Coding Agent Index evaluation, Copilot Fusion Max scored 60 points, which is only 2 points lower than Claude Code running Opus5.5Max, but the average API cost is about half, the Token consumption is halved, and the task completion speed is about 40% faster. This feature is fully available in all GitHub Copilot plans, covering Visual Studio Code, CLI, applications and JetBrains IDE. There is no additional orchestration fee and is only billed based on the actual call of the model. A 30% model fee discount is provided until November 30, 2026. Enterprise administrators can control available models and set them as the default experience. In addition, the system also supports experimental local model routing, such as calling Microsoft MAI Code1.1Flash on Windows devices. Microsoft simultaneously launched the Copilot harness developed by the GitHub team, which has enabled a variety of products including Copilot Studio and VS Code, and is open to developers through Microsoft Foundry. Multi-model intelligent orchestration is becoming an important evolution direction of AI programming tools, and is expected to achieve a better balance between performance and cost. via AI News (author: AI Base)