LinkedIn post size.

LinkedIn post sizes and limits, checked against official pages

Every row names the surface it applies to. “Not stated” means the official page we read did not publish that value; it is not a gap filled from a third-party guide. All pages returned HTTP 200 and were read on 2026-10-29 between 14:25 and 14:26 UTC.

The general media page is broader than the photo-post page. Its 100 MB and 36-megapixel figures do not replace the 5 MB line on the photo-post page, and this page does not explain why the two documents differ. Both are listed with their own context instead of being merged into one rule.

The 1200 × 627 minimum belongs to the website link preview, and 1200 × 628 belongs to the horizontal Single Image Ads layout; the ad page has separate square and vertical sections with their own values. Neither figure is a photo-post rule. A photo at 1200 × 627 is 1.91:1, which sits inside the published 3:1 to 4:5 range and above 552 × 276 px; the photo-post page does not label that size invalid. What the photo page does say is narrower: if the ratio is larger than the maximum ratio, the image is centered and cropped to fit that maximum. The page read does not state a crop mechanism for ratios outside the range on the tall side.

Multi-photo posts allow up to 20 images and emphasize the first one. This tool handles one image at a time; it does not simulate a multi-photo layout, and it does not claim to reproduce LinkedIn's own crop or rendering.

Crop one LinkedIn post image in your browser

Pick one of the three tool canvases above, then choose Fill to cover the canvas or Fit to keep the whole image with a background colour. The 2× option doubles both canvas dimensions; those doubled pixels are a tool extension, not a LinkedIn recommendation or maximum.

Choose an image to begin.

No image loaded.

Download imageCopy image

What the encoded file shows

• Choose an image to read back its actual pixels, ratio, bytes and format.

After every encode the file is decoded again and these values are read from that decoded copy. The checks compare it with the published minimum, ratio range and our 5,000,000-byte reading of “5 MB”; they do not predict whether LinkedIn will accept, crop or recompress it.

What our own crop test measured

We built a 1620 × 1080 px test image from code: an 90 px grid, a dark frame, four corner markers and three lines of bitmap text, including a line set 28 px tall. The source, six exports, six 320-pixel-wide previews, their SHA-256 hashes and the measurement file are in linkedin-post-samples/results.json. Run make_samples.py with Python 3 to rebuild every file; it uses only the standard library, with no clock, randomness or network access.

The script applies the same geometry as the browser tool. It uses nearest-neighbour sampling so the run is byte-stable; the browser uses its own resampler, so the byte counts below describe these files, not a browser download. All six exports were measured after PNG encoding by re-reading the PNG header and the file length.

How the crop and pad areas were calculated

Let the source be 1620 × 1080 px and the canvas be 1080 × H px. Fill scales up until the image covers the canvas and then keeps one centred rectangle. Fit scales down until the whole image fits and fills the remainder with the background colour. The 2× option doubles both canvas dimensions and leaves each share unchanged, because every scale factor doubles with the canvas.

Fill: scale = max(1080 / 1620, H / 1080); kept = (1080 / scale) × (H / scale); removed share = 1 − kept area / 1,749,600

Fit: scale = min(1080 / 1620, H / 1080); drawn = (1620 × scale) × (1080 × scale); background share = 1 − drawn area / (1080 × H)

The removed share and the background share match in every pair: 25.00% at 2:1, 33.33% at 1:1 and 46.67% at 4:5. That is geometry, not a coincidence. When the source is wider than the canvas ratio, Fill discards the same fraction of the source that Fit replaces with background. The square and portrait crops also remove the outer corner markers; the Fit versions keep all four and add white bands instead.

What the 320-pixel-wide previews show

320 px is our local observation width. It is not a LinkedIn display size and it is not measured on LinkedIn. At that width the 28 px label lands between 4.15 px and 10.37 px tall depending on the canvas and mode. The 4:5 Fill export is the only one that keeps it above 10 px, because that crop keeps 864 of the 1620 source columns and then scales them up by 1.25. Every Fit export uses the same 0.667 scale, so the label measures 5.53 px at 320 px wide, and the 2:1 Fit export is smaller still at 4.15 px.

In the files themselves, the 4:5 Fill preview still shows “SMALL TEXT 28 PX” as separate strokes, while the 2:1 Fit preview turns it into a grey band. The corner markers survive every Fit run and disappear from the square and portrait Fill runs. Those observations come from opening the six PNG previews, not from an upload test.

Limits of this test

The shares can be checked by hand: 1 − (1620 × 810) / (1620 × 1080) = 0.25 at 2:1, 1 − (1080 × 1080) / (1620 × 1080) = 0.3333 at 1:1 and 1 − (864 × 1080) / (1620 × 1080) = 0.4667 at 4:5. Fit reaches the same three shares because the drawn area ratio is identical.

This is a controlled local test, not a LinkedIn upload test. It uses one synthetic source, one focus point (50%), PNG only, and nearest-neighbour resampling. PNG is used because it is lossless, so the byte counts respond to geometry and pixel colour content (the generator writes RGB truecolor) rather than JPG quantisation; a JPG export of the same canvas will differ in bytes; a browser's smoother resampler and a JPG encoder will produce different pixels and bytes. No file was uploaded, so the test says nothing about LinkedIn's server-side processing. It also does not score “quality”: it measures geometry, encoded size and the legibility of one 28 px label after a local downscale.

What it does support is narrower and repeatable: on a 1620 × 1080 source, Fill at the 4:5 canvas keeps 53.33% of the source area, Fit at 2:1 keeps 75.00% of the canvas as image, and a small text line loses about half its apparent height when the export is viewed at 320 px wide. Those numbers can be recomputed from the formulas above and re-measured by rerunning the script.

Why your LinkedIn post image can still look soft

A local crop throws pixels away. The 4:5 Fill run above removed 46.67% of the source area before encoding, so any small detail in the discarded columns cannot appear in the export. Fit keeps every source pixel but makes the image smaller inside the canvas, which reduces the size of fine text instead of removing it. Neither operation can add detail that the source never had.

Your original file sets the ceiling. A 600 px wide photo exported to 1080 px wide is enlarged; the canvas is bigger, but the detail is not. A 2160 px wide photo exported to a 1080 px canvas is reduced, which is usually where text and thin lines start to merge. Compare the original, the actual export and a 320 px wide local preview before uploading; that comparison is local evidence, not a prediction of what LinkedIn will show.

What LinkedIn does after upload is not published in the pages we read. We do not claim to know its compression settings, its encoder quality or any sharpening percentage, and we do not repeat unsupported claims about Featured images or company logos. The published rules this page can check are the 5 MB line, the 552 × 276 px minimum, the 1080 px recommended width and the 3:1 to 4:5 range.

Questions

Sources

• Share photos on LinkedIn — photo-post size limit, minimum pixels, recommended width, aspect range and the 20-image count. Read 2026-10-29 14:25 UTC.
• Media file types supported on LinkedIn — broader image formats and the 100 MB / 36-megapixel table. Read 2026-10-29 14:25 UTC.
• Image specifications for your LinkedIn Pages and Career Page — website link preview values. Read 2026-10-29 14:25 UTC.
• Single Image Ads — ad file types, the 5 MB limit and the horizontal/landscape 1200 × 628 px recommendation. Read 2026-10-29 14:25 UTC.
• Differences in image size and third-party links for organic posts and Sponsored Content — separate organic and Sponsored link-preview recommendations. Read 2026-10-29 14:26 UTC.

All five pages were publicly readable without signing in on the date above. Where a page did not state a value, the table says Not stated; no value was imported from a community post or another site.

Other image tools

Working on a profile cover or a Page cover instead? Use the LinkedIn banner size tool; its limits are separate from feed photos.

Making a feed image for another network? Open the Instagram post size tool or return to the ThumbGlance preview tool.