A viral post analyzer for 100,000 X posts (X)

Play the video in the post below

Movez pulled 100,000 viral X posts through a few X APIs and sent them to Jev in one parallel pass, with 14 typed yes/no questions per post: does the hook open a loop, is there a number in the first line, is the proof real or claimed. Each answer comes back with a confidence score. He makes the distinction explicit: viral analysis is a classification job, not writing.

The full batch finished in 20.4 seconds for $0.67. As a comparison, he ran the same corpus on Claude Opus 5 with the same clock, and it got through 214 posts for $0.98. He puts that at roughly 680 times the per-post cost of Jev, and says a full Opus pass over the 100,000 posts would have cost $458.

What the classifications turned up: 1,220 posts broke into the top 1%, a 1.22% baseline rate. Posts with a superlative claim hit 2.34%, a 1.92x lift; contrarian takes hit 1.59%, a 1.31x lift; launch or tool-drop posts hit 1.46%, a 1.19x lift. Numbered lists, the format he says everyone writes, came in at 0.55%, below baseline, which makes the most-used hook the least viral one in his data.

Open on X