API Release Notes

Launched context caching for Gemini 2.0 Flash.

Model updates:

• Released veo-2.0-generate-001, a GA text- and image-to-video model. See the Veo docs.
• Released gemini-2.0-flash-live-001, a public preview Live API model with billing enabled, featuring:
  • Session Resumption: Server-side session state storage (up to 24 hours) with reconnection handles.
  • Longer Sessions via Context Compression: Sliding window context compression to prevent abrupt terminations.
  • Graceful Disconnect Notification: GoAway server message before connection closes.
  • Configurable Voice Activity Detection (VAD): Sensitivity levels or manual turn control via activityStart/activityEnd events.
  • Configurable Interruption Handling: Control whether user input interrupts the model.
  • Configurable Turn Coverage and Media Resolution.
  • Expanded Voice & Language Options: Two new voices, 30 new languages, configurable output language.
  • Text Streaming: Incremental text responses.
  • Token Usage Reporting: Detailed token counts in usageMetadata.

Released gemini-2.5-pro-preview-03-25, a public preview Gemini 2.5 Pro version with billing enabled. The experimental version gemini-2.5-pro-exp-03-25 remains available on the free tier.

Released gemini-2.5-pro-exp-03-25, a public experimental Gemini model with thinking mode always on by default. See Gemini 2.5 Pro Experimental.

Model updates:

• Launched an experimental Gemini 2.0 Flash image generation and editing model.
• Released gemma-3-27b-it, available on AI Studio and through the Gemini API, as part of the Gemma 3 launch.

API updates:

• Added support for YouTube URLs as a media source.
• Added support for inline video of less than 20MB.

Released the Google Gen AI SDK for TypeScript and JavaScript to public preview.

Released gemini-embedding-exp-03-07, an experimental Gemini-based embeddings model in public preview.

Added support for Search as a tool to gemini-2.0-pro-exp-02-05, an experimental model based on Gemini 2.0 Pro.

Released gemini-2.0-flash-lite, a generally available (GA) version of Gemini 2.0 Flash-Lite, optimized for speed, scale, and cost efficiency.

Added support for additional regions (Kosovo, Greenland, and Faroe Islands) for both AI Studio and the API.

Gemini 1.0 Pro is no longer supported. See Gemini models for the list of supported models.

Updates on the OpenAI libraries compatibility.

• Released imagen-3.0-generate-002, a generally available (GA) version of Imagen 3 in the Gemini API.
• Released the Google Gen AI SDK for Java for public preview.

• Released gemini-2.0-flash-001, a GA version of Gemini 2.0 Flash supporting text-only output.
• Released gemini-2.0-pro-exp-02-05, an experimental public preview of Gemini 2.0 Pro.
• Released gemini-2.0-flash-lite-preview-02-05, an experimental preview model optimized for cost efficiency.
• Added file input and graph output support to code execution.
• Released the Google Gen AI SDK for Python to general availability (GA).

Released gemini-2.0-flash-thinking-exp-01-21, the latest preview version of the Gemini 2.0 Flash Thinking Model.

Released Gemini 2.0 Flash Thinking Mode for public preview. Thinking Mode is a test-time compute model that lets you see the model's thought process while it generates a response, producing responses with stronger reasoning capabilities. See the overview page.

Released Gemini 2.0 Flash Experimental for public preview. Features include:

• Twice as fast as Gemini 1.5 Pro
• Bidirectional streaming with the Live API
• Multimodal response generation (text, images, speech)
• Built-in tool use with multi-turn reasoning (code execution, Search, function calling, and more)

See the overview page.

• Released gemini-exp-1121, a more powerful experimental Gemini API model.
• Updated gemini-1.5-flash-latest and gemini-1.5-flash aliases to use gemini-1.5-flash-002. Note: gemini-1.5-flash-002 supports top_k values between 1 and 41 (exclusive); values greater than 40 will be changed to 40.

Released gemini-exp-1114, a powerful experimental Gemini API model.

Added support for Gemini in the OpenAI libraries / REST API.