GitHub Introduces ‘Premium Requests’ for Copilot AI Usage Across Advanced Models
Copilot Pro users will receive 300 monthly premium requests, while Copilot Business and Enterprise tiers will receive 300 and 1,000 requests
GitHub has unveiled a new system called ‘premium requests’ for its AI coding assistant, Copilot, which introduces rate limits when developers use AI models other than OpenAI’s GPT-4o, the default model. This change applies to newer integrations such as Anthropic’s Claude 3.7 Sonnet and others.
While unlimited usage of GPT-4o remains free, more advanced tasks—like agentic coding and multi-file edits—done with alternative models will count as premium requests.
Starting May 5, Copilot Pro users will receive 300 monthly premium requests, while Copilot Business and Enterprise tiers will receive 300 and 1,000 requests, respectively, beginning between May 12–19.
Users can purchase extra requests at $0.04 each, or opt for Copilot Pro+ at $39/month, which includes 1,500 premium requests.
GitHub has also rolled out Agent Mode in Visual Studio Code to general availability, allowing users to select from AI models such as Claude 3.5 & 3.7 Sonnet, Gemini 2.0 Flash, and GPT-4o. New features like Code Review Agent and Next Edit Suggestions are also now available.
Additionally, GitHub launched an open-source local MCP server, enabling integration of any compatible LLM with GitHub workflows like repository search and pull request creation.
