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

2 points by linzhangrun ↗ HN

2 comments

[ 2.3 ms ] story [ 18.0 ms ] thread
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.
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.