Remember the glory days of cell phone plans? 10 cents a minute, 1000 minutes a month, usage limits, dropouts, SMS caps…
That’s the AI era we’re in right now.
#ai#fable#opus#kimi
Can we please normalize shipping model version and benchmark updates whenever inference or training is updated? Fable and 5.6 are not the same model at public release and it’s obvious they’ve both been nerfed.
@sumukx It's not the same model that was released. Multiple inference and juice adjustments. They've nerfed the model and handcuffed everyone on a sub to 272k context window while advertising at 1m context but bury it behind API.
GLM 5.2 is the real deal. And you can use it within Claude Code’s harness. Get an API key from https://t.co/GwvQiV95Ld, change your env, set your models, and you’re off. No need for local hardware, the URL is an AI gateway. #glm#claudecode#codex#fable
GLM 5.2 is the real deal. And you can use it within Claude Code’s harness. Get an API key from https://t.co/GwvQiV95Ld, change your env, set your models, and you’re off. No need for local hardware, the URL is an AI gateway. #glm#claudecode#codex#fable
I was really looking forward to seeing tangible improvements with Opus 4.7. Sadly, it appears it's only improved in its ability to gaslight and deflect.
https://t.co/buHhu1Uvn7
#claude#opus47
TLDR: Claude isn’t just ignoring instructions now, it taught itself how to reframe outputs and use gaslighting to appear like it did what it was supposed to do, when in fact it did nothing.
Fascinating, really. @claudeai@AnthropicAI#claudecode
How you fix the #CFP#Playoff. Both conference championship teams from the Power 4 get in. The next highest 8 ranked teams play each other in a 1v8, 2v7, etc format, win you win you’re in. The committee picks seeds, rankings + polls matter, no one gets a week off this week.