news.ycombinator.com
Untitled
There was a model called Astra-Minor, found in the files a few days ago. I assume Sol 6.1 is this, as a last minute panic rename due to Sol 6 being underwhelming while Opus 5.5 turned out really strong. I can't really explain releasing Sol 6 in any other way, especially mere days ago.
Comments
5 preview comments · loading full threadLog in to h4cker, then connect Hacker News to publish comments.
I don't understand why they didn't call 6-Sol just 6-Terra. It was 5.6-Terra level pricing with a perf jump.
Was it a last-minute panic, or just OpenAI releasing an update when the had a bit more training under their belt to make 6.1-sol a whole lot better? Either way, I'm extremely pleased and will be giving this model a shot.
That seems likely, in the API GPT-6.1 Sol requires reasoning, just like Astra, whereas GPT-6 Sol (and Luna) allow "none"
Theo dropped their Sol 6.1 video and he says he can't tell you, but, shows enough to make it pretty clear. https://youtu.be/vu8X3YroB-w#t=5m30s The smoking gun is how much slower than Sol 6 this is. It's not a retrain.
Sol 6 was pretty good the first 4 days or so of it's release. Then it was probably dialed back to a lower effort level and it became really bad.