Skip to content

Ask HN: Does anyone else find GPT-5.6 Sol in Codex slow?

5 pointslinzhangrun2 comments
On HN

Comments

It feels like all models are slow(er). I've switched back to 5.5 Medium and it also took 12 minutes to do a small-ish refactor across 8 files. I guess that happens when you onboard nearly a million users per day while also removing the 5h usage limit.

It’s a little bit slower compared to 5.5 xhigh for me but I don’t have the numbers to justify it, just “feels”. 5.6 sol on xhigh seems to dig a bit deeper and find more edge cases so it takes longer than predecessors.

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.