File Upload, Vision, and OCR in KoalaChat

KoalaChat isn't a text-only chat. Drop files in and the chat reads them: extracting text from images, parsing tables out of spreadsheets, understanding code files, and answering questions that span multiple attachments at once.

Supported formats

• Images: JPEG, PNG, GIF, WebP
• Documents: PDF, TXT, CSV, Markdown, DOCX, XLSX, PPTX, Jupyter notebooks (.ipynb)
• Code: PY, JS, TS, JSX, TSX, JSON, HTML, CSS, XML, YAML, SH, RB, GO, JAVA, C, C++, H, RUST, SQL
• Academic: LaTeX, BibTeX, ODT, TEX

Attachment limits

Up to 10 attachments per message in regular KoalaChat. Brand DNA chats cap at 5 to stay inside agent context limits.

OCR (image-to-text)

Scanned receipts, screenshots, photos of whiteboards, image-based PDFs: KoalaChat runs OCR on them and feeds the extracted text into the conversation. Pricing:

• Free monthly OCR page allowance (varies by plan).
• Overage billed at 1 credit per 10 pages beyond the allowance.

Document tokens

Long PDFs and DOCX files are tokenized for the model. Pricing:

• Free monthly document-token allowance.
• Overage billed at 1 credit per 20,000 tokens beyond the allowance.

What you can do with it

• Paste a screenshot of a chart and ask the chat to interpret it.
• Drop a PDF of a research paper and ask for a summary + key findings.
• Upload a spreadsheet and ask for a SQL query that would produce the same view.
• Send a code file and ask for a code review or refactor suggestions.
• Upload an invoice (image) and have it parsed into structured fields.
• Send a whiteboard photo from a meeting and get a transcribed action list.

Attachment persistence

Files are saved as active_attachments on the message they were attached to. When you revisit an old chat, the file metadata is preserved (filename, type, category) so the conversation reads correctly. The raw bytes may not be retained indefinitely; the conversation continues to reference the extracted text.