Files, images & video

Harness can read the files you give it and make new ones: images, short videos and documents.

Giving Harness files

Attach files to a message in the Harness window: Word, PowerPoint, Excel, PDF, text, Markdown, CSV, JSON, and PNG, JPEG or WebP images. You can attach up to 5 files per message, up to 15 MB each. Documents are read on your device, and Harness reads long ones page by page as it needs them.

Documents

Ask for a document and Harness writes it as a real file you can download: a Word document, a PowerPoint deck, an Excel workbook with working formulas, or any plain-text file such as CSV, JSON, Markdown or code. Ask for changes and it edits the same file.

Images

Ask Harness to draw something, or to change an image you attached, and it makes the image in the conversation. It uses Nano Banana 2 by default. In the private privacy setting it uses a private image model, Z-Image Turbo, instead.

Video

Harness can make a short video from a description, or bring an image to life. It uses Veo 3.1 Fast. Video is available in the anonymous privacy setting only. If Harness thinks a video would help when you did not ask for one, it offers it on a card first, with Render or Not now.

Availability and cost

Making documents, images and video is a Resident feature. Images and video are paid from your managed credit, like other managed usage. None of them run in the private e2e setting or when you are signed out. Reading the files you attach is available on every plan. See Plans.