Hi everyone,
We’re evaluating gemini-3-flash-preview for a client-facing system and need to assess migration risk before committing. Could someone from the Gemini team clarify the following:
Deprecation/retirement date — Is there a fixed shutdown date for this preview model, or is it “available until further notice”?
Migration path to GA — Once this preview is retired, will there be a stable (GA) version with the same behavior/API signature, or will we need to change the model ID entirely?
Advance notice policy — How much advance notice (days/weeks) does Google typically give before retiring a preview model? Is this documented anywhere specific to Preview-tier models vs GA models?
Preview vs GA differences — Does this preview version have different rate limits, quota, or lack of uptime SLA compared to GA models?
Pinning behavior — If we hardcode this model ID in production, will requests fail after retirement, or auto-redirect to a newer version?
I checked ai.google.dev/gemini-api/docs/models but didn’t find a specific retirement date listed for this preview. Any pointers to official documentation, or confirmation from the team, would be greatly appreciated.
Thanks!