Choose Your Edit Prediction Provider

Zed has expanded its functionality to support multiple edit prediction providers, allowing users to choose between Zeta, Mercury Coder, Sweep, Ollama, Codestral, and GitHub Copilot's Next Edit Suggestions. This enhancement enables users to select a prediction model that best aligns with their coding style, aiming to improve coding efficiency by anticipating changes, suggesting syntax completion, and optimizing code editing. Zed's edit prediction system now features a pluggable provider architecture, simplifying the integration of new prediction models by focusing on model-specific components like API calls and response handling. Zeta remains the default model, but users can experiment with other options like Mercury Coder, which uses a diffusion architecture for parallel output refinement, or Sweep, which offers rapid suggestions using a custom diff format. Ollama allows for local execution of open-weight models, ensuring code privacy. Users can configure their preferred provider in the Settings Editor, and Zed encourages experimentation to find the best fit based on language, project size, and editing style.