React Performance Review: test results

Verdicts

Date

2026-10-29

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

Right on all 3 cases, read by hand. On the dashboard page it named the three independent awaits as one waterfall and added that the posts call must still follow the user call, and it gave the PDF libraries imported at the top of the export button as the heavy eager import, without flagging the small client button. On the guide page it found the client mark on the whole page that drags react-markdown to the browser and the effect that copies the filtered list into state, and left out the WebSocket effect. On the clean orders page 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 pairs of findings as Sonnet on the dashboard and on the guide page, with the same safe look-alikes left out, and answered exactly No findings on the clean orders page. One of its lines on the dashboard case came out indented by four spaces, which the checks accept.

Note

Light test: three cases (two planted, one clean), one run per model; no comparison without the skill. The code was written by us; each planted case carries a safe look-alike (a dependent await, a small client button, an effect that syncs a socket), and neither planted case has a comment pointing at the defect. One check was wrongly strict and is disclosed: Sonnet's dashboard answer mentioned the posts call in a parenthesis to say it must stay after the user call, and the first version of the check rejected any waterfall line that named it; the check was narrowed to a waterfall line that names only the posts call, and the six saved answers were re-evaluated without a new model call. Facts come from the public React and Next.js documentation, read on 29 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.