Gemini API now reads legacy Office and OpenDocument files

Alongside it, tool calls can hand back audio and video, the embedding-2 preview string is retired, and Omni 1.1 Flash reroutes Dec 3.

Nowline Nov 7 7:00 AM banner

Top AI stories from the last hour

Top AI stories from the last hour

Copy markdown

  • Feed it PPT, ODP, ODS — no conversion step

    The Gemini API now accepts PowerPoint 97-2003 (.PPT), OpenDocument Presentation (.ODP), and OpenDocument Spreadsheet (.ODS), and legacy Word (.DOC) now gets page-image processing too. DOC/PPT/ODP render as page images; ODS passes the first sheet's data and formulas. DOCX, PPTX and XLSX handling is unchanged — so you can drop a client's legacy Office files straight into a prompt instead of wiring up a conversion step first.

  • Your tools can now hand back audio and video

    Function responses in both the GenerateContent and Interactions APIs now accept audio, video, and Office/OpenDocument blocks plus fileData URIs — not just text. Video parts take a STATIC or AGENTIC media-processing flag. Practically: a tool that fetches a clip can return it for the model to watch and reason over in the same turn, instead of you transcribing it first.

  • gemini-embedding-2-preview is retired — swap one string

    As of Nov 6, requests to gemini-embedding-2-preview route to gemini-embedding-2 at GA rates. Google says the schema, pricing, and embedding space are identical and no re-indexing is needed — so it's a one-line model-string change, but do it before the preview name stops resolving.

  • Deprecation clock: Omni 1.1 Flash reroutes Dec 3

    From December 3, requests to gemini-omni-1.1-flash will route to gemini-omni-1.2-flash at 1.2 rates. If you pinned 1.1 for predictable output or cost, test against 1.2 now — paid-tier limits doubled and video output dropped 30% to $12 per 1M tokens.