Gemini AI v2 is completely backward compatible. The only change you may have to be aware of is that configuration settings no longer error when giving a nonexistent field to account for future API changes.

Major Changes:

• Library is now in TypeScript for type safety
• File Uploads are now dynamically optimized (automatically uses File API for videos/audio, and inline data for images)
• MIME types are now detected by the file-type library, so they should be much more reliable
• New message format (directly pass an array of buffers/strings into the message parameter)
• Ability to change API version
• New default model is now gemini-1.5-flash-latest, and optimizations made for Gemini 1.5 models
• Streaming is now done through SSE, so it should be much more reliable
• Chat is now completely dependent on Gemini's ask() method, ensuring parity
• Many improvements have been made under-the-hood from the rewrite!