FILE PHOTO: GitHub has announced a new system for GitHub Copilot called ‘premium requests’ so rate limits are imposed when using other AI models rather than the base model. (Thomas Dohmke, CEO)
| Photo Credit: THB

GitHub has announced a new system for GitHub Copilot called ‘premium requests’ so rate limits are imposed when using other AI models rather than the base model, which is OpenAI’s GPT-4o. This includes new models that have been added, like Anthropic’s 3.7 Sonnet. 

As unlimited queries with GPT-4o stay free, any tasks like “agentic” coding and multi-file edits with other AI models will be considered as premium requests. 

Starting May 5, Copilot Pro will include 300 monthly premium requests while Copilot Business and Copilot Enterprise customers will get 300 and 1,000 monthly premium requests respectively from between May 12 and May 19. 

These customers can also buy additional premium requests at $0.04 per request or upgrade to GitHub’s Copilot Pro+, which costs $39 every month and allows 1,500 premium requests from a range of AI models. 

GitHub has also made ‘Agent Mode’ in Visual Studio Code generally available, and enhanced Model Context Protocol (MCP) support. Agent Mode allows users to choose from between AI models including Anthropic’s Claude 3.5 and 3.7 Sonnet, Google’s Gemini 2.0 Flash and OpenAI’s GPT-4o. 

A couple of other features – Code Review Agent and Next Edit Suggestions – have also been made generally available now. 

GitHub further released a new open-source and local GitHub MCP server so any LLM that supports MCP can be integrated with GitHub functionalities like searching repositories and creating pull requests.