Google · Text models · Developers

Gemini API reads PPT, ODP and ODS files, and tools can return media

Google's Gemini API now accepts PPT, ODP and ODS input, renders DOC files as page images, and lets function responses include audio, video and documents.

Google’s Gemini API release notes, in an entry dated November 5, 2026, widen the file types the models can read and let tool results carry media. A separate entry on November 6 retires the Gemini Embedding 2 preview name.

What changed

  • More office formats: the API now accepts PPT, ODP (OpenDocument slides) and ODS (OpenDocument spreadsheets) as input, and adds page-image processing for DOC files.
  • How they are read: Google says DOC, PPT and ODP files are rendered as page images. For ODS, the model receives the data and formulas from the first sheet only. Processing of DOCX, DOTX, ODT, PPTX, XLSX and XLS is unchanged.
  • Richer function responses: when your code answers a model’s function call, the response can now include audio, video and Office or OpenDocument files, plus fileData URIs, in the GenerateContent API. In the Interactions API, function results accept document, audio and video blocks; Google says video blocks are in v1beta and v1alpha.
  • Embedding name change (November 6): gemini-embedding-2-preview is deprecated and requests now route to gemini-embedding-2 at its rates. Google says the schema, pricing and embedding space are unchanged, so no re-indexing is needed, but you should update the model string.

What to do

  1. If a multi-sheet ODS file matters, put the sheet you need first or export the other sheets separately, since only the first sheet is read.
  2. Slide decks are read as page images, so ask for page numbers in answers and check charts and small text yourself. Our long document review templates show how.
  3. If your agent tools fetch files or recordings, you can now hand them straight back to the model instead of converting them to text first. Keep approval rules in place; see our agent permissions checklist.

The same day Google made Gemini Omni 1.2 Flash generally available. For other recent changes, see the Gemini 2.5 deprecation, the Gemini tool page and Google’s release notes.

Source: Google AI Studio release notes. We summarise in our own words; check the source for the full details.

More AI news

stay in the loop

New guides, prompts and AI model news by email