Migrating off gemini-2.0-flash

gemini-2.0-flash was a mainstay of Google's Gemini free tier — the default reach-for model for high-volume, cost-sensitive work, and a common first model for indie developers building on Gemini. Google retired it on the date shown above.

Provider	Google
Shutdown date	2026-06-01
Replacement	`gemini-3.5-flash`
Official source	Google deprecations page

Why it was retired

Google moves the Gemini flash line forward quickly, and retires older Flash versions once a newer one covers the same low-cost, low-latency niche. After the retirement date above, calls that name gemini-2.0-flash no longer resolve.

Because Flash is the tier people pick precisely so they don't have to think about it, a retirement here tends to hit background jobs and free-tier prototypes that nobody is actively watching.

What to move to

The replacement above is the current Flash-class model — same role in the lineup: the cheap, fast default. Moving to it keeps you on the tier Flash was designed for rather than forcing a jump to a pricier model.

If you were on the free tier, re-check the current quota and rate limits for the new model; Google adjusts these between versions, and the numbers you relied on for gemini-2.0-flash are not guaranteed to carry over.

Migration notes

Gemini's API is broadly compatible across Flash versions, but two things reward a second look: multimodal handling (how images and other inputs are passed) and safety-setting defaults, both of which have shifted between Gemini generations and can change output on inputs that worked before.

Update the id wherever it lives — application code, the google-generativeai client setup, notebooks, and any config or environment variable — since the bare string is what stopped resolving.