I've been seeing a lot of hype around kimi k 2.7 and glm 5.2, with benchmarks putting them around frontier models for a fraction of the cost. But I want to know if anyone has actually replaced opus or gpt with them yet or if its all just talk
@SullyOmarr This assumes 1 of 2 things. Either the project is built for technical people or the project is executed poorly. In both cases I agree it’s a stupid project idea that’s bound to fail
@danielcberk Honestly I think it’s the exact opposite. PMs feel like the last person to go because now they can just leverage ai rather than their direct reports. In the post ai world non technical people aren’t the first to be replaced
@Tech2Wild Local models are definitely the safety net for when the bubble bursts but currently they arent the smartest choice in my opinion. Most users aren’t coming close to spending the thousands of dollars that hardware costs on api calls.
Fable is just as I expected. Obviously it's extremely powerful but the cost just makes it unusable for 99% of users. Gpt 5.5 and Opus 4.8 were already good enough for almost any task and I just don't see most users needing that performance boost at that cost.
I've been building a fitness app mostly because when i first started lifting i had no idea what i was doing. Getting into lifting is so overwhelming with all the noise you hear. There's no one right way to workout but people online act like there is and overcomplicate it.