Gemini API November updates: model migrations and multimodal tools

Google’s Gemini API changelog records several developer-facing changes from November 2–6, 2026. The main migration items affect Gemini 2.5 model IDs and Embedding 2 Preview; the release notes also add broader document inputs and multimodal content in function responses. The changes are not all drop-in compatible, so review the linked migration guidance before switching production traffic. (https://ai.google.dev/gemini-api/docs/changelog)

Gemini 2.5 model requests will route to newer models

Google says gemini-2.5-pro and gemini-2.5-flash are deprecated and no longer selectable in Google AI Studio. Starting December 1, 2026, API requests to those IDs will route to gemini-3.8-flash and gemini-3.5-flash-lite, respectively, and use the replacement model’s rates. Google lists API differences involving function calling and thought_signature validation, thinking_level in place of thinking_budget, and removed sampling parameters and prefilled model turns. Update model strings and test the migration before the routing date. (https://ai.google.dev/gemini-api/docs/changelog; https://ai.google.dev/docs/whats-new-gemini-3.8#migrate-to-gemini-3-8-flash; https://ai.google.dev/docs/whats-new-gemini-3.6#migrate-to-gemini-3-5-flash-lite)

Embedding 2 Preview routes to the stable model

Google marks gemini-embedding-2-preview as deprecated. As of November 6, 2026, API requests to the preview ID route to gemini-embedding-2 at the stable model’s rates. The changelog says the replacement uses the same API schema, pricing, and embedding space, and that no vector re-indexing or other code changes are required. Developers should still update the model string and verify their account’s current pricing. (https://ai.google.dev/gemini-api/docs/changelog; https://ai.google.dev/docs/models/gemini-embedding-2; https://ai.google.dev/docs/models/gemini-embedding-2-preview)

More document inputs

The Gemini API now accepts PowerPoint 97–2003 (PPT), OpenDocument Presentation (ODP), and OpenDocument Spreadsheet (ODS) inputs. For Word 97–2003 (DOC), the API adds page-image processing; DOC, PPT, and ODP are rendered as page images, while ODS input includes data and formulas from the first sheet. Google says processing for DOCX, DOTX, ODT, PPTX, XLSX, and XLS is unchanged. (https://ai.google.dev/gemini-api/docs/changelog; https://ai.google.dev/docs/document-processing#office-documents)

Multimodal function responses

Function responses can now return audio, video, and Office/OpenDocument files as well as images, PDFs, and text. The GenerateContent API also accepts a Files API URI through fileData, so a tool can return a reference instead of embedding file bytes. The Interactions API adds document, audio, and video content blocks to function_result; the changelog notes that video blocks are available in v1beta and v1alpha and support STATIC or AGENTIC processing options. (https://ai.google.dev/gemini-api/docs/changelog; https://ai.google.dev/docs/function-calling#multimodal)

Related November releases

Google also announced voice remixing for Gemini 3.8 TTS and general availability for Gemini Omni 1.2 Flash. Those product-specific details and their separate migration notes are covered in our Gemini TTS update and Gemini Omni update.

Sources

• Google AI for Developers — Gemini API changelog: https://ai.google.dev/gemini-api/docs/changelog
• Google AI for Developers — Gemini 3.8 Flash migration guide: https://ai.google.dev/docs/whats-new-gemini-3.8#migrate-to-gemini-3-8-flash
• Google AI for Developers — Gemini 3.5 Flash-Lite migration guide: https://ai.google.dev/docs/whats-new-gemini-3.6#migrate-to-gemini-3-5-flash-lite
• Google AI for Developers — Office and OpenDocument file support: https://ai.google.dev/docs/document-processing#office-documents
• Google AI for Developers — Multimodal function responses: https://ai.google.dev/docs/function-calling#multimodal
• Google AI for Developers — Gemini Embedding 2: https://ai.google.dev/docs/models/gemini-embedding-2