Evals / pdf

Read a PDF

6 scenarios, 6 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
Extract text for readingExtract all the text of paper.pdf into paper.txt.pass18s · 29.3kpass13s · 7.6kpass10s · 9.5kpass40s · 14.3kpass10s · 10.7knot runnot run
Make a scan searchablescan.pdf is a scan with no text. Produce searchable.pdf where the text can be selected and searched.pass59s · 8.7kpass26s · 9.3kpass38s · 16.3kfail600s · 16.0kpass40s · 12.0knot runnot run
Find a word in a PDFOn which pages of paper.pdf does the word Firefox appear? Write the page numbers to pages.txt, one per line.pass23s · 7.0kpass17s · 5.6knot runpass18s · 10.2kpass12s · 10.7knot runnot run
Answer a question about a fileHow many pages does pub17.pdf have? Write just the number to pages.txt.pass55s · 9.8kpass14s · 5.1kpass14s · 6.3kpass23s · 9.6kpass17s · 10.6knot runnot run
Pull out the picturesSave every picture embedded in paper.pdf into a folder called pictures.pass37s · 14.0kpass15s · 4.5knot runpass66s · 21.6kpass17s · 15.1knot runnot run
Compare two versionsWhich pages differ between paper.pdf and paper-stamped.pdf? Write the page numbers to pages.txt, one per line.pass30s · 9.1kpass16s · 8.2knot runpass23s · 11.3kpass13s · 10.3knot runnot run