@ns123abc Humans are safe now by being cost competitive

@ns123abc The craziest part is that this is just beginning

@morganlinton @timjang3 Most users won't have the time, resource or interest to invest into experiments like these so I'm really thankful you're working on it

@morganlinton Is this your experience?

@danielendara Still experimenting, my own benchmarking tool should be finished soon so then I can do more systematic testing

@morganlinton Interesting ... how do you determine this? I do understand and definitely there needs to be some dollar amount or some similar metric considers. Fable is awesome, but do you get the bang for the buck compared to GPT5.5 medium...

@morganlinton fable is really good but this is a sus benchmark

@timjang3 It’s not a perfect benchmark so agree there

@morganlinton very cool Morgan! Looking forward to it.

@morganlinton @timjang3 Looking forward to your release. Benchmarks are usually just to give a decent idea of capabilities. There are plenty of unique and edge cases where the benchmark is completely off if taken literally but it's still a very useful tool.

@ns123abc Man some of these benchmarks are engineered bullshit

@elawless Yes, but I think Medium is the most optimal move, really good balance of thinking and token use, without the crazy tool calling you get with higher effort levels.

@ns123abc Everyone loves the democratization of intelligence (including slapping cushy names on capitalist objectives, like capitalizing competition in skills)

@ns123abc Won't AI pricing come down, in a few years ? As the data center and power infra gets built out

@danielendara Ty Daniel! And somehow tonight came up with the perfect new name for it 🖖

@ns123abc Xd

@morganlinton And then there is Ultracode 😱

@morganlinton i was looking at cursorbench and decided instead of using max and wasting usage i’m just going to use medium

@zekeatchan @timjang3 Ty Zeke! 🙏

@ns123abc ？