Gemini Deprecated Models and Endpoint Retirement Schedule

The next Gemini API shutdowns land on October 22, 2026, when Google turns off the three Veo 3.1 preview models and gemini-omni-flash-preview. The deep-research-pro-preview-12-2025 agent follows on October 23, 2026, and five speech and live audio previews shut down on November 17, 2026. At Layer3Labs, we build AI workflow systems for businesses, and model shutdowns are a routine maintenance task in those builds. Any code still calling one of those model IDs after its date stops getting responses.

Earlier generation endpoints have already reached their cutoff dates. Google shut down gemini-2.0-flash and gemini-2.0-flash-lite on June 1, 2026, and names gemini-3.6-flash and gemini-3.1-flash-lite as their replacements. A shut-down model is completely turned off, so its endpoint no longer answers Application Programming Interface (API) calls.

Every date below comes from Google's Gemini API deprecations page, where a listed date is the earliest a model can be shut down. Verify your production model strings against the official Google deprecations documentation before scheduled shutdown dates arrive.

Gemini Models Shutting Down Next

Ten preview models face retirement between October 22, 2026, and November 17, 2026, in the Gemini API. The earliest cluster of retirements arrives on October 22, 2026, removing three Veo 3.1 video generation previews alongside gemini-omni-flash-preview. Google designates gemini-omni-1.1-flash as the official replacement for all four retiring media snapshots. Teams using those endpoints must update their model configuration strings to maintain continuous service.

The managed agent endpoint deep-research-pro-preview-12-2025 shuts down on October 23, 2026, with deep-research-preview-04-2026 serving as its direct successor. The subsequent wave lands on November 17, 2026, retiring five audio and real-time endpoints. This group includes preview Text-to-Speech (TTS) models from the Gemini 2.5 and Gemini 3.1 generations, as well as two live audio preview models.

Speech synthesis workflows targeting the retiring TTS previews must migrate to either gemini-3.8-flash-tts or gemini-3.8-flash-lite-tts. Real-time conversational pipelines using gemini-3.1-flash-live-preview or gemini-2.5-flash-native-audio-preview-12-2025 must transition to gemini-3.8-live. Confirm your active model identifiers in codebases against the Google deprecations documentation before these November cutoffs.

• veo-3.1-generate-preview (released October 15, 2025): shuts down October 22, 2026 → replace with gemini-omni-1.1-flash
• veo-3.1-fast-generate-preview (released October 15, 2025): shuts down October 22, 2026 → replace with gemini-omni-1.1-flash
• veo-3.1-lite-generate-preview (released March 31, 2026): shuts down October 22, 2026 → replace with gemini-omni-1.1-flash
• gemini-omni-flash-preview (released June 30, 2026): shuts down October 22, 2026 → replace with gemini-omni-1.1-flash
• deep-research-pro-preview-12-2025 (released December 11, 2025): shuts down October 23, 2026 → replace with deep-research-preview-04-2026
• gemini-3.1-flash-tts-preview (released February 26, 2026): shuts down November 17, 2026 → replace with gemini-3.8-flash-tts or gemini-3.8-flash-lite-tts
• gemini-2.5-flash-preview-tts (released May 20, 2025): shuts down November 17, 2026 → replace with gemini-3.8-flash-tts or gemini-3.8-flash-lite-tts
• gemini-2.5-pro-preview-tts (released May 20, 2025): shuts down November 17, 2026 → replace with gemini-3.8-flash-tts or gemini-3.8-flash-lite-tts
• gemini-3.1-flash-live-preview (released March 11, 2026): shuts down November 17, 2026 → replace with gemini-3.8-live
• gemini-2.5-flash-native-audio-preview-12-2025 (released December 12, 2025): shuts down November 17, 2026 → replace with gemini-3.8-live

Scheduled Retirements from 2027 to 2028

Google has scheduled three Gemini API models for retirement across 2027 and 2028. The earliest of these extended dates occurs on March 15, 2027, when the multimodal image endpoint gemini-2.5-flash-image shuts down. Google lists gemini-3.1-flash-lite-image as its replacement.

The model gemini-3.1-flash-lite is scheduled to shut down on May 7, 2027, exactly one year after its release. Applications running on this endpoint should prepare to migrate to gemini-3.5-flash-lite. The furthest scheduled retirement is gemini-embedding-001, which remains supported until May 14, 2028, before giving way to gemini-embedding-2.

The 2027 and 2028 dates leave time to run your existing prompts against gemini-3.1-flash-lite-image, gemini-3.5-flash-lite and gemini-embedding-2 before switching traffic. Pinned model strings in infrastructure configuration should be scheduled for review well ahead of the cutover dates. Always cross-reference your configuration files with the Google deprecations documentation during quarterly dependency audits.

• gemini-2.5-flash-image (released October 2, 2025): shuts down March 15, 2027 → replace with gemini-3.1-flash-lite-image
• gemini-3.1-flash-lite (released May 7, 2026): shuts down May 7, 2027 → replace with gemini-3.5-flash-lite
• gemini-embedding-001 (released July 14, 2025): shuts down May 14, 2028 → replace with gemini-embedding-2

Completed Gemini API Model Retirements

More than twenty Gemini, Imagen, and Veo model snapshots were already shut down between late 2025 and October 2026. The most prominent retirement occurred on June 1, 2026, when Google shut down gemini-2.0-flash, gemini-2.0-flash-001, gemini-2.0-flash-lite, and gemini-2.0-flash-lite-001. Requests directed to those model strings fail because the endpoints are completely inactive.

Several Gemini 2.5 and Gemini 3 previews shut down earlier. Google retired three snapshots of gemini-2.5-pro-preview on December 2, 2025, recommending gemini-3.1-pro-preview as their replacement. The specialized computer interaction model gemini-2.5-computer-use-preview-10-2025 reached its shutdown on July 28, 2026, routing developers to gemini-3.8-flash.

Image and video models were also retired. Google retired three Imagen 4.0 endpoints on August 17, 2026, replacing them with gemini-nano-banana-2.1. Earlier video generation snapshots, including veo-2.0-generate-001 and two Veo 3.0 variants, shut down on June 30, 2026. Search your code for each retired ID below.

• gemini-2.0-flash and gemini-2.0-flash-001: shut down June 1, 2026 → replaced by gemini-3.6-flash
• gemini-2.0-flash-lite and gemini-2.0-flash-lite-001: shut down June 1, 2026 → replaced by gemini-3.1-flash-lite
• gemini-2.5-pro-preview-03-25, -05-06, and -06-05: shut down December 2, 2025 → replaced by gemini-3.1-pro-preview
• gemini-2.5-flash-preview-09-25: shut down February 17, 2026 → replaced by gemini-3.6-flash
• gemini-3-pro-preview: shut down March 9, 2026 → replaced by gemini-3.1-pro-preview
• gemini-2.5-flash-lite-preview-09-2025: shut down March 31, 2026 → replaced by gemini-3.1-flash-lite
• gemini-3.1-flash-lite-preview: shut down May 25, 2026 → replaced by gemini-3.1-flash-lite
• gemini-3.1-flash-image-preview and gemini-3-pro-image-preview: shut down June 25, 2026 → replaced by gemini-nano-banana-2.1 or gemini-3-pro-image
• veo-3.0-generate-001, veo-3.0-fast-generate-001, and veo-2.0-generate-001: shut down June 30, 2026 → replaced by Veo 3.1 previews (which themselves shut down October 22, 2026, with gemini-omni-1.1-flash as their replacement) or General Availability (GA) models on Gemini Enterprise Agent Platform
• gemini-2.5-computer-use-preview-10-2025: shut down July 28, 2026 → replaced by gemini-3.8-flash
• imagen-4.0-generate-001, imagen-4.0-ultra-generate-001, and imagen-4.0-fast-generate-001: shut down August 17, 2026 → replaced by gemini-nano-banana-2.1
• gemini-robotics-er-1.6-preview: shut down August 31, 2026 → replaced by gemini-robotics-er-2-preview
• antigravity-preview-05-2026: shut down October 5, 2026 → replaced by antigravity-preview-09-2026
• text-embedding-004 (shut down January 14, 2026), embedding-001 (shut down October 30, 2025), and embedding-2-preview (shut down August 10, 2026) → replaced by gemini-embedding-2
• gemini-2.0-flash-live-001 and gemini-live-2.5-flash-preview: shut down December 9, 2025 → replaced by gemini-3.8-live

Gemini API Deprecation Policies and Endpoint Behavior

Google treats published shutdown dates as the earliest possible retirement date for any Gemini API model. The official policy notes that Google will communicate the exact shutdown date to users with advance notice. Once an endpoint reaches its confirmed shutdown date, Google turns it off completely and it is no longer available to handle calls.

The deprecation documentation does not specify a distinct HTTP error code returned when sending calls to a shut-down endpoint. Because the endpoint is turned off, client SDK and direct REST requests to it fail. If Gemini calls start failing, compare the model string in your code against the retired IDs listed above.

One exception covers Gemini 2.5 models such as gemini-2.5-pro, gemini-2.5-flash and gemini-2.5-flash-lite. Google states that Gemini 2.5 models remain available to users who used them before and continue to be served until further notice. Dated Gemini 2.5 previews, such as gemini-2.5-pro-preview-tts, still carry their own shutdown dates. Teams can confirm retirement rules directly on the Google deprecations documentation.

Current Gemini Models with No Announced Shutdown

At least fourteen current Gemini, embedding, audio, and specialized models have no shutdown date announced as of October 15, 2026. They include text, speech, embedding and image models. gemini-3.8-flash, released September 2, 2026, is the replacement Google names for gemini-3.7-flash and gemini-2.5-computer-use-preview-10-2025.

Developers building voice and audio applications can use gemini-3.8-live and gemini-3.8-flash-tts without impending retirement dates. In the image and multimodal generation category, Google released gemini-omni-1.1-flash on August 27, 2026, and gemini-nano-banana-2.1 on October 6, 2026. Neither model has a scheduled cutoff date.

Two interim models, gemini-3.7-flash and gemini-3.5-flash, show shutdown as not applicable in Google documentation, but list designated replacements. Google points gemini-3.7-flash workloads toward gemini-3.8-flash, and gemini-3.5-flash workloads toward gemini-3.6-flash. When starting fresh implementations, choose endpoints with clear long-term support windows and verify their status in the Google deprecations documentation.

• gemini-3.8-flash: released September 2, 2026 (no shutdown announced)
• gemini-3.8-live: replacement for gemini-3.1-flash-live-preview and gemini-2.5-flash-native-audio-preview-12-2025 (no shutdown announced)
• gemini-3.8-flash-tts: replacement for the retiring TTS previews (no shutdown announced)
• gemini-3.6-flash: replacement for gemini-2.0-flash and gemini-3.5-flash (no shutdown announced)
• gemini-3.5-flash-lite: replacement for gemini-3.1-flash-lite (no shutdown announced)
• gemini-3.1-pro-preview: replacement for gemini-3-pro-preview and the gemini-2.5-pro previews (no shutdown announced)
• gemini-3-flash-preview: preview model (no shutdown announced)
• gemini-2.5-pro: no shutdown announced; served to past users until further notice
• gemini-2.5-flash: no shutdown announced; served to past users until further notice
• gemini-2.5-flash-lite: no shutdown announced; served to past users until further notice
• gemini-embedding-2: replacement for gemini-embedding-001 and text-embedding-004 (no shutdown announced)
• gemini-nano-banana-2.1: released October 6, 2026 (no shutdown announced)
• gemini-omni-1.1-flash: released August 27, 2026 (no shutdown announced)
• lyria-3.5: no shutdown announced
• gemini-3.7-flash: shutdown listed as N/A → recommended replacement is gemini-3.8-flash
• gemini-3.5-flash: shutdown listed as N/A → recommended replacement is gemini-3.6-flash

Model Migration Steps and Platform Scope

Migrating off a deprecated Gemini model requires swapping the model identifier string in your API requests and testing the payload against the replacement endpoint. Because parameter support and response formatting can vary between model generations, validate your schema definitions before updating production traffic. Run your existing prompts against the replacement model and compare the outputs before the shutdown date.

These dates apply only to the Gemini API at ai.google.dev. Google Cloud Vertex AI manages enterprise infrastructure separately and publishes its own distinct deprecation schedule. A Vertex AI model can have a different cutoff date. Gemini CLI is a separate case because it is a tool, not an API model: Google replaced it with Antigravity CLI for consumer accounts on June 18, 2026, as covered in our Google Antigravity CLI guide.

If your code also calls OpenAI or Anthropic models, the OpenAI deprecated models and Anthropic model deprecations pages list their shutdown dates. To audit your Gemini endpoints against current releases, compare your configurations with the Google deprecations documentation.