GPT-6.1 Sol

“6.1 sol really does seem like something now”

“models are a bit faster since the reset”

“Peor que el 5.6”

“flying through the 5 hour now on sol 6.1”

“I use GPT-6.1 Sol through the Codex CLI as the reviewer, high effort, on the uncommitted diff before every commit and once more against main before a PR.”

“I asked it to do ABC, it said okay I'll do ABC, then only did BC”

“it will not do as you ask. It will do whatever it thinks is a better method.”

“can barely keep up with my banked resets expiration dates now”

“lazy model who does the easiest solution always so you need to iterate with it multiple times”

“Really, I also feel it's extremely stupid. It completely ignores the original project and just writes a whole new set of code. I'm really speechless”

“my 5 hour usage was gone in 1 prompt and the prompt wasnt even half finished”

“I've let 6.1 run successfully overnight and still have plenty of use left in the morning wow!”

“thats why they cut our usage in half”

“Told it to modify a function implementation, it changed the interface, and all the calls were also changed”

“They should release a slow mode, then even 20 bucks can go all out”

“Telling it to remove ads from several software tools and unlock paid features, it did it without asking a single question”

“terrible with anything vision related”

“Both Astra and 6.1-Sol are pretty comfortably ahead of Opus 5.5 in these types of evaluations”

“The hacking capabilities are restricted to death.”

“Civil servant mentality, likes to pass the buck, unwilling to take responsibility, and drags out work by doing safe tasks.”