Grok 4.7 is a disappointing release.

I already forgot about it, lol We need you to stream Today, its gonna be a VERY FUN stream :)

There's no way I can stream and also film the videos I want to the level of detail I want. Gonna keep digging into the models offline and get videos out ASAP :)

Lets gooo, cant wait to watch your stream 🫡 My trusted source :) Side idea: Maybe you can stream to us how you try models & stuff? Would still be fun to watch

AI models are hitting a weird point where “smarter” doesn’t always feel better. If users prefer the older model, the benchmark doesn’t matter

Yeah I hope they just go straight to Grok 5 and skip .8 & .9, straight up 10T for fun

Theo the audio is a bit messed up I think. It's not synced.

Wait until you see the insane fucking level of regression in both 6 Sol and Opus 5.5. It’s genuinely so disappointing

ChatGPT apphas a daybreak toggle option so we can use some models with these reduced guardrails for cyber. could we add this into t3code for accounts that have daybreak access please 🙏

I agree I'm quite disappointed too.

You better watch out for Teslas on the road now.

ANT released AGI and you are talking about grok ??

Coul you do your fish slop test on mimo v2.6 pro?

Same experience

felt this running agents overnight. 4.5 i could just leave on and it did the work. 4.7 bills me like double for the same job and still fumbles frontend. cost per finished task is the only number i care about anymore

TALK ABOUT OPUS 5.5

u have a shit load of videos for today ig 😭

if it gets slower and burns more tokens, a small benchmark bump won’t save it. speed matters when you’re using it all day.

You have obviously no idea what you are talking about. Its not AI but SI. Thank you for your attention to this matter.

i guess dario and sam also agree with u this time. just not elon

Maybe they'll do a 4.7.5 model release or something to address some of these issues?

yep its terrible with gpt 6 astra, sol, luna, opus 5.5 and fable 5.1 as well as sonnet 5.5 and haiku 5.5 soon we dont need this

we are so busy with opus 5,5 and SOL6 - why this release now

it reasons a lot a lot, comes up with correct findings actually. Good for code reviews. Not okay for writing code.

Looks even worse after Sol6...

Price/performance is only half the story—tool reliability and context retention matter more once a model is in a real workflow. A week of usage data would be fascinating.

I dont mind, i just saw it in the model picker and replaced the 4.6 xhigh with 4.7 xhigh. Its like no noticeable difference in smaller targetted tasks, but if its more correct then great. I already blacklisted grok from larger architectural changes

They need to improve that's for sure

Good video, I agree with your findings and observations

was lowkey expected to be a dissapointment after Elon's comments and the delays

It was expected. Its out of the party. It remembers me when Google plus was rushing to compete with facebook. LOL

Calling the release disappointing in public is the rare AI launch ritual that might actually get the next one fixed.

Grok was yesterday 😂

bro, but we care about opus, sol, and luna right now

4.7? sounds more like 3.9 with marketing.