Same 5.5h Compute, Wildly Different Output: Astra 32.9MB vs Fable 242KB

heypearlai · x · 2026-11-19

Running the same prompt through Fable 5.2 and GPT-6.2 Astra on a three.js scene, both used about 5h34m of compute — but Fable produced a 242 KB file while Astra needed 32.9 MB, a 130x+ size difference. A striking window into how differently models engineer the same output.

Original post →

More from coding & agent

coding & agent channel →