@amasad If you look at the latest benchmark results the logical conclusion is that grok 4.5 does a lot of the work. If it is non trivial. Or am I missing something
If you’re not using grok 4.5 as a daily driver it’s negligence at this point. Yeah, I still brainstorm and review and plan with Sol+Fable+grok if the task is extremely multifaceted and the tentacles of consequences and blast radius is unknown and stakes are very high. But for any other task. It’s sheer financial negligence. Tell me where I’m wrong
If you’re not using grok 4.5 as a daily driver it’s negligence at this point. Yeah, I still brainstorm and review and plan with Sol+Fable+grok if the task is extremely multifaceted and the tentacles of consequences and blast radius is unknown and stakes are very high. But for any other task. It’s sheer financial negligence. Tell me where I’m wrong
@MattWalshBlog lots of patients feeling the sting today lol. it's not easy to justify to themselves that they have wasted years and an astounding amount of money on pure bullshit
@ori_pomerantz that will never work if the program is 50k, 100k+ loc. agents can churn out that amount of code in a week, even fairly responsibly if they are controlled by the things @unclebobmartin mentions. you will never ship if you have to read all that code
@SECNAV fucking hell. how can you be in the navy in any capacity at all, nevermind a director, and be this fucking stupid? i wouldn't even tell a stranger what i'm working on even if it were mildly corporate sensitive. my god
@imjaredz@grok what do you expect are some medium term consequences of cracking these kinds of math problems? Are they just curiosae, interesting? Or do they translate to real world value somehow