Gemini API changelog

Released across the Interactions and GenerateContent APIsGA

Stateless, zero-dependency command-line client for coding agents, scripts and CI pipelines, built on Interactions APIPreview

New paid-tier projects store requests by default (store=true); existing projects keep their setting

Model built for complex coding, agentic, math and science tasksGA

Deprecated models route to newer versions; old Deep Research agent shuts down October 23, 2026Deprecation

Image generation and editing model succeeds gemini-3.1-flash-image; adds 1:8, 8:1 and other wide aspect ratiosDeprecation

New text-to-speech models and a Voices endpoint with 150+ prebuilt and custom voicesGA

Access limited to users who have previously used them; not deprecated, served until further notice

Replaces the 05-2026 version, which shuts down October 5, 2026; local built-in tools changedDeprecation

Two audio-to-audio models for real-time voice apps in Live API, one with extended background reasoningGA

Full-length song generation with text and image inputs and 44.1 kHz stereo audioGA

Flash model aimed at long-horizon software engineering, autonomous agents and enterprise workflowsGA

Released for Gemini 3.7 Flash, 3.6 Flash and 3.5 Flash-Lite, using up to 88% fewer tokens

Adds video extension, first/last frame interpolation and 360p to 4K resolution controlDeprecation

Two speech-to-text models: batch Transcribe and streaming Transcribe Live over WebSocketsGA

Model for coding and agents, with introductory pricing through December 31, 2026GA

Two embodied reasoning endpoints added; ER 1.6 preview shuts down August 31, 2026Deprecation

Stable releases of both models; temperature, top_p and top_k parameters deprecatedDeprecation

Logs for supported Interactions API calls are viewable in the dashboard

Video generation and conversational editing model, 3-10 second videos at 720p; Flash Lite Image reaches GAPreview

Computer Use tool adds intent-based actions, browser, mobile and desktop support, safety policies and prompt injection dPreview

Streaming supported for speech generation via streamGenerateContent and the Interactions API

Imagen shuts down August 17, 2026; Veo 2 and 3.0 shut down June 30, 2026Deprecation

Four model versions shut down; use gemini-3.5-flash or gemini-3.1-flash-liteDeprecation

gemini-3.1-flash-image and gemini-3-pro-image released; video-to-image added; previews shut down June 25, 2026Deprecation

Model shut down; gemini-3.1-flash-lite replaces itDeprecation

gemini-3.5-flash is GA and behind gemini-flash-latest; Managed Agents and Antigravity Agent in public previewGA

Stable release replaces preview model, which shuts down May 25, 2026Deprecation

New schema default May 26; legacy schema removed June 8Deprecation

Multimodal search over images with gemini-embedding-2, plus media_id and page_numbers in grounding metadata

Event-driven webhooks replace polling for Batch API and long-running operationsGA

Model shut down; use gemini-robotics-er-1.6-preview insteadDeprecation

Embedding model released as generally availableGA

New versions add collaborative planning, visualization, MCP server integration and File SearchPreview

Cost-efficient, expressive, steerable text to speech model releasedPreview

Robotics model adds instrument reading and better spatial reasoning; 1.5-preview shuts down April 30, 2026Deprecation

gemma-4-26b-a4b-it and gemma-4-31b-it released in AI Studio and Gemini APIGA

Adds Flex and Priority tiers for trading off cost and latencyPrice

Cost-efficient video generation model launched; gemini-2.5-flash-lite-preview-09-2025 shut downDeprecation

Audio-to-audio model for real-time dialogue and voice-first applications, via Live APIPreview

lyria-3-clip-preview makes 30-second clips; lyria-3-pro-preview makes full-length songs, 48kHz stereoPreview

Prepay and Postpay billing plans rolled out; existing accounts may be affectedPrice

Built-in tools can be combined with function calling; Google Maps grounding for Gemini 3

Revamped Usage Tiers and Billing Account spend caps introducedPrice

Project-level spend caps added to billing in AI Studio

First multimodal embedding model handles text, image, video, audio and PDF; flash-lite preview shuts down March 31, 2026Deprecation

Model shut down; gemini-3-pro-preview now points to gemini-3.1-pro-preview

First Flash-Lite model in the Gemini 3 seriesPreview

Nano Banana 2 image model launched; Gemini 3 Pro Preview shuts down March 9, 2026Deprecation

New Gemini 3 series model, plus customtools endpoint for mixed bash and custom toolsPreview

Models shut down June 1, 2026, including the -001 versionsDeprecation

gemini-2.5-flash-preview-09-25 and two imagen-4.0 preview models shut down

Computer Use tool in preview for gemini-3-pro-preview and gemini-3-flash-preview

gemini-pro-latest and gemini-flash-latest aliases now point to Gemini 3 previews

Three preview models to shut down February 17, 2026; gemini-2.5-flash-image-preview shut downDeprecation

text-embedding-004 model shut down

Veo adds 4k output resolutions and wider portrait video support

Gemini API adds model lifecycle stages and deprecation timelines

Gemini API accepts Cloud Storage and pre-signed URLs as input, file limit now 100MB