Zed dev: Gemini 4 Neon solves in one round what 3 frontier models couldn't crack in 9

vivnat · x · 2026-11-19

A Zed developer shares a striking data point: he runs Opus 5.5, Sol 6.1, and Gemini 4 Carbon as a mixture-of-experts 'cross-review round' on Zed's hardest algorithmic challenges (difficult Presburger entailment issues). Two problems resisted all three frontier models even after 6 and 9 full rounds of cross-review spanning hours. Gemini 4 Neon then solved both problems in a single round — 'Neon is not Carbon lite, it's way more powerful than Carbon.'

Original post →

More from coding & agent

coding & agent channel →