Back to Blog
AI Models June 3, 2026 5 min read

Google Retires All Gemini 2.0 Models as the 3.x Family Takes Over

Google shut down four Gemini 2.0 production models on June 1, completing the platform's transition to Gemini 3.x. Developers still calling 2.0-flash endpoints are now hitting deprecation errors.

Google Retires All Gemini 2.0 Models as the 3.x Family Takes Over

Google retired all four Gemini 2.0 production models on June 1, completing the platform’s shift to the Gemini 3.x family. The discontinued models are gemini-2.0-flash, gemini-2.0-flash-001, gemini-2.0-flash-lite, and gemini-2.0-flash-lite-001. Any application still pointing at these endpoints is now receiving deprecation errors.

The official migration path points to gemini-3.5-flash for production agentic and coding workloads, and gemini-3.1-flash-lite for latency-sensitive or cost-optimized applications. Google’s API documentation positions Gemini 3.5 Flash as the default general-purpose production tier going forward.

The retirements arrive five days after Google pushed two new image-generation models into general availability on May 28. gemini-3.1-flash-image and gemini-3-pro-image are now production-ready, both supporting video-to-image generation — a capability that had been preview-only for months. The Pro variant targets high-fidelity generation for design and media workflows; the Flash variant is optimized for throughput at lower cost per call.

Two preview models enter their own deprecation countdown simultaneously. gemini-3.1-flash-image-preview and gemini-3-pro-image-preview will be shut down on June 25. Developers relying on these previews for prototyping have three weeks to migrate to their GA equivalents.

The pace of version turnover is worth noting. Gemini 2.0 Flash launched in early 2025 as Google’s fastest production model and quickly became one of the highest-volume models in the API market by call count. Retiring the entire 2.0 Flash family within roughly a year of launch reflects how decisively the 3.x generation outpaced it on performance and pricing.

For teams building on the Gemini API, the migration checklist is straightforward: audit any model string hard-coded as gemini-2.0-flash*, update to gemini-3.5-flash or gemini-3.1-flash-lite, and validate outputs against your evaluation suite before deploying. The response format and tool-use API surface are compatible across the 2.0 and 3.x Flash families, so migration typically amounts to a model string change.

Apps that haven’t migrated yet are not just missing out on newer capabilities — they’re actively broken. This is a hard cutover, not a soft deprecation with a grace period.

Sources

Google Gemini AI models API