GPT-6.1 Sol ran in circles for days on ts-rust, sparking regression complaints

karmay007 · x · 2026-09-30

Reacting to @theo's progress post on his Rust project ts-rust, karmay007 highlights that GPT-6.1 Sol was thrown at the task for several days and "ran in circles and made no progress." He concludes that 5.6 Sol outperforms both 6.0 and 6.1 Sol, quipping that "we are officially REGRESSING" — a notable counterpoint to the newest model's positioning.

Original post →

More from coding & agent

coding & agent channel →