Evals / image

Size and compression

3 scenarios, 3 run by an agent. Each run shows its result, time and new tokens.

ScenarioCodexwithout deftCodexwith deft, skill installedCodexwith deft, skill as agents.mdClaude Opus 5.5without deftClaude Opus 5.5with deft, skill installedClaude Codewithout deftClaude Codewith deft
Resize a large photoMake a 1200 pixel wide version of earth.jpg called web.jpg.fail8s · 2.9kpass12s · 4.5kpass10s · 9.4kpass16s · 11.0kpass8s · 9.8knot runnot run
Hit a size limit without resizingearth.jpg is 6MB. Get it under 500KB without changing its dimensions, as small.jpg.pass18s · 4.5kpass14s · 5.3kpass17s · 10.4kpass24s · 10.7kpass13s · 11.1knot runnot run
Thumbnails for a mixed folderCreate 200 pixel wide thumbnails of every image in the photos folder, in a folder called thumbs.pass14s · 6.1kpass19s · 6.2kpass15s · 10.6kpass25s · 11.5kpass18s · 11.9knot runnot run