kimi k3.1 pro came out this week and everyone keeps comparing it to opus 5.5. so we ran it through our writing benchmark to see who actually writes better same job for both. 10 youtube scripts, 5 drafts each, 4 ai judges scoring every draft without knowing which model wrote it honestly kimi did better than we expected. on the opening hook it's a tie, 83.4 for kimi vs 82.9 for opus, and each one won the hook on 5 of the 10 scripts. on facts and depth it's a tie too, 88.2 vs 88.4 the gap is the voice. opus sounds like a real person talking (82.5 vs 77.0) and its scripts read smoother out loud (83.2 vs 77.8). kimi's drafts are correct but they sound more like a model wrote them that one gap is enough for opus to win all 10 scripts. opus is #1 on the whole board, kimi lands at #7 of 83 models but kimi costs a quarter of the price. $0.25 a script vs $1.00, so 1,000 scripts is about $251 vs $1,003 so if every word matters, opus. if you write a lot and you're going to fix the voice yourself anyway, kimi gets you most of the way for way less one of our 4 judges is a claude model, so we checked again without it. opus still came out ahead full side by side with every score is here benchmark.towardsai.com/comp…

Anthropic let Claude run a real shop in their office it lost money, made up a payment account, and told a customer it was standing by the fridge in a blazer and red tie the whole story is wild 🧵

cut your ai writing bill by 98% without the ai slop the best ai writers cost $1,000 or more for every 1,000 youtube scripts. claude opus 5.5, our #1 at 85.8 out of 100, comes to about $1,003 but the cheap ones sound like a robot. setups under $30 per 1,000 scripts had a median