Platform Text Limits: test results

Verdicts

Date

2026-10-19

Strong · claude-sonnet-5-5 (Claude Code alias "sonnet")

Right on all 3 cases, read by hand. On the search ad it reported headline 2 at 35 of 30 and description 2 at 114 of 90, with shortened texts of 25 and 85 characters, and left out the headline of exactly 30 characters. On the App Store listing it counted the keywords as 110 bytes and left out the subtitle of exactly 30. On the X post it answered exactly No findings.

Weak · claude-haiku-5-5 (Claude Code alias "haiku")

Right on all 3 cases, read by hand. It gave the same two counts for the search ad with shortened texts of 25 and 90 characters, the byte count of 110 for the keywords, and exactly No findings for the X post. The longest shortened text sat exactly on the limit of 90, which is allowed. Both models kept the numbers, the prices and the sixty-day returns promise of the original description, and neither invented a claim.

Note

Light test: three cases (two planted, one clean), one run per model; no comparison without the skill. The three inputs were written by us; each planted one carries a safe look-alike and neither has a comment pointing at the defect. Facts come from the public pages named in the notes, read on 19 October 2026; there are no owner-measured facts.

What was not measured

• Models other than the two named above were not run.
• Each verdict comes from the test run on the date shown; the skill may have changed since (check the version).
• Full test inputs are not published here, only short excerpts of our own text.
• Results on your own texts, languages and domains can differ.