ChomsGP describes recurrent problems with the Gemini Pro model in GitHub Copilot, contrasting it with more stable alternatives, and raises concerns about its reliability and value.

Summary

In this community post, ChomsGP outlines persistent technical difficulties encountered when using the Gemini Pro model within both GitHub Copilot (including the web-based chat and VSCode agent mode). Unlike other models, Gemini Pro frequently exhibits significant lag, hanging issues, and frequent request failures, rendering it nearly unusable in their workflow.

Main Points

  • Model Comparison: While other AI models perform adequately, Gemini Pro consistently fails, especially during premium requests.
  • Usability Issues: The user reports severe lagging and errors, particularly when using VSCode in agent mode, and also with reading pull requests in GitHub chat—tasks it previously completed successfully.
  • Reliability Concerns: These issues lead the author to question the value of premium features, as the error rate and instability make the experience subpar compared to other providers previously used by the author.
  • Long Context Advantage: Despite these problems, the long context capability of Gemini Pro is seen as a strong point; however, its unreliability overshadows this benefit.
  • Context: The author mentions dissatisfaction with other providers (such as Cursor, due to pricing changes), driving their interest in Gemini Pro despite its shortcomings.

Community Impact

ChomsGP’s feedback reflects a broader challenge for developers relying on premium AI integrations. Persistent technical failures can significantly hinder productivity and diminish the value proposition of advanced features.

Conclusion

The post serves as a cautionary note for prospective Gemini Pro users on GitHub Copilot, emphasizing the importance of reliability in premium AI development tools.

This post appeared first on Reddit Github Copilot. Read the entire article here