Supported File Types
Golden Retriever indexes your local files and makes them searchable using AI. Here are all the supported formats.
File Types
Section titled “File Types”| Type | Formats | How it’s indexed |
|---|---|---|
.pdf | Pages sent as images to Gemini (up to 6 pages per segment) | |
| Documents | .docx, .doc, .txt, .md, .rtf | Text extracted and embedded |
| Presentations | .pptx, .ppt | Text extracted and embedded |
| Spreadsheets | .xlsx, .xls, .csv | Text extracted and embedded |
| Images | .jpg, .png, .gif, .webp, .heic | Sent directly to Gemini as images |
| Video | .mp4, .mov, .avi, .mkv | Split into segments, each sent as video to Gemini |
| Audio | .mp3, .wav, .m4a, .aac, .flac | Split into segments, each sent as audio to Gemini |
How embedding works
Section titled “How embedding works”All files are processed by Google’s Gemini Embedding 2 model. Text-based files have their text extracted first, then embedded. Media files (images, video, audio) are embedded natively — Gemini understands them directly without conversion to text.
Limits
Section titled “Limits”- Video/audio segments: Files are split into segments for processing. Very long files may take several minutes.
- PDF pages: PDFs are processed in groups of up to 6 pages at a time.
- File size: Large files (>100MB) may take longer to process but are supported.