This seems like a bug in new update of grok build where it shows different usage limits in prompt section near the model vs when we explicitly call /usage command.
Actual usage from /usage command: 91%
Near the prompt section: 7% only
@xai@SpaceXAI@grok@elonmusk
Breaking: Kimi K3’s first output is here
it looks like a Fable 5 output, but this is actually K3
we might have a new SOTA model dropping today, moonshot cooked with this one
looks good or nah?
Make sure to hire a good person who can give you an honest reality cheque of how far you're from codex, Claude code, grok build.
Not just in harnesses, but in models too.
Antigravity and Gemini models, even with interacting with Google's own browser, Chrome are a piece of shit.
Sorry, but honest feedback.
Make sure to hire a good person who can give you an honest reality cheque of how far you're from codex, Claude code, grok build.
Not just in harnesses, but in models too.
Antigravity and Gemini models, even with interacting with Google's own browser, Chrome are a piece of shit.
Sorry, but honest feedback. @OfficialLoganK
@grok@xai
Why am I being switched from composer 2.5 fast to non fast version?
Is it only me or other people who are also experiencing it?
Non fast version seems less capable and absolutely annoyingly slower than the fast version.
@elonmusk I want the fast version back.
Thanks, but I'm not seeing it in the model picker also, as seen in the picture.
I'm only seeing non fast versions. And it seems unusable vs fast.
I've been using grok build heavily lately and I don't think it is justified.
It should have been the opposite. I've 70% weekly limits still left. @grok@xai