gemini-2.0-flash-thinking-exp-01-21

Google (Gemini API) retired gemini-2.0-flash-thinking-exp-01-21 on 2 December 2025.

ProviderGoogle (Gemini API)
Statusretired
Shutdown date
Deprecated on
Replacementnone named
Sourcehttps://ai.google.dev/gemini-api/docs/changelog
Noterelease notes shutdown announcement

Move off it now

  1. Find every call to it in your code (Python, JS/TS, Go, Java, configs, notebooks), with nothing to install:
    uvx llmshift scan
  2. Turn your real traffic into tests, so you compare on what your users send:
    llmshift snapshot logs/ --model "gemini-2.0-flash-thinking-exp-01-21"
  3. See exactly what changes on the new model: formats, facts, tone, tool calls and settings it rejects:
    llmshift compare --to <new model>
  4. Swap the model in your code and open a pull request with the results:
    llmshift upgrade --from "gemini-2.0-flash-thinking-exp-01-21" --apply --compare --pr

Then llmshift init --github keeps the check on every pull request. LLMShift is free and open source (MIT) and runs on your machine.

The error you get now

Calls to a model after its shutdown fail. The answer usually reads (the wording varies by SDK and endpoint):

404 NOT_FOUND: models/gemini-2.0-flash-thinking-exp-01-21 is not found for API version v1beta, or is not supported for generateContent.

Other Google (Gemini API) retirements

Wrong date or replacement? Tell us on GitHub, with a link to the provider's page.

Lifecycle data updated 2026-11-14, collected from the providers' official deprecation pages. Dates can move: check the source before you plan around one. Visits are counted with GoatCounter, which sets no cookies.