LinkedIn Document ads: 100 MB, 300 pages, PDF rules and slides

A LinkedIn Document ad file can't exceed 100 MB or 300 pages, and you can upload up to five documents. Supported types are PPT, PPTX, DOC, DOCX and PDF, but a Document ad with a lead generation objective only supports PDFs. Those limits are from LinkedIn's Document ads advertising specifications page, read on 2026-10-03.

The page also sets a word limit of 1 million. If you build the document from AI slide art, the limit that bites is file size, not words: full-bleed images at print resolution add up fast.

Document specs and text limits

Two rules affect how you build the pages. PDFs with multiple layers must be flattened or merged, and PDFs with pages of different sizes must be fit to the same page size. Videos and animations in the file display as static images only, and hyperlinks are clickable only after the document is downloaded and must be secure.

What the page does not cover

The specs page lists no cover image size, thumbnail rule or gate-page requirement, so this post does not claim any. It also does not say how many pages a viewer can see before a gate, only the 300-page ceiling. Keep a Document ad to a page count you would actually read; a tight 10 to 20 page deck is a design choice, not a LinkedIn rule.

Also, a Document ad is not a Thought Leader ad. LinkedIn's Thought leader ads page lists documents as ineligible for promotion at this time, so a member's document post cannot be boosted that way. See which posts qualify.

What Sume makes for the pages, and what it does not

Sume makes the slide art, not the PDF. The Image API lists openai/gpt-image-2.5, which takes a prompt, up to 16 image references and background: auto|transparent|opaque, with quality from low to max. For a 16:9 slide, set image_size to 1792x1008: GPT custom sizes need both edges to be multiples of 16, a maximum edge of 3840, an aspect ratio of at most 3:1, and between 655,360 and 8,294,400 pixels, and 1792 x 1008 passes all four.

Generate one image per slide, keep the type simple (the docs advise raising quality for dense text and packaging), then place the images in your deck tool and export a PDF. Sume has no PDF, PPT or Word output, so flattening layers and checking the 100 MB limit happen in that tool. Each image job is async, so poll it as described in Jobs and results.

curl -X POST https://api.sume.com/v1/images \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: deck-slide-01" \
  -d '{
    "model": "openai/gpt-image-2.5",
    "prompt": "Slide 1 of a B2B report deck: clean title slide, large headline Q3 Findings, flat shapes, plenty of margin",
    "image_size": "1792x1008",
    "quality": "high",
    "output_format": "png"
  }'

Keep it under 100 MB

Use output_format: "jpeg" for photographic slides and PNG only where you need flat color or transparency. Export the PDF at screen resolution, not print. If the file is still over 100 MB, lower the image resolution or cut pages before you export again.