Partially blind, how? I'm guessing you are referring to Chinese models not saying anything negative about China or its government? Or Taiwan?
You've clearly never had a conversation with chatgpt or claude about race, religion, gender, or any number of other topics that give weird answers?
@benjitaylor They are very versatile!
it's why i've spent the last few days putting together grokbot.studio, a different take on what every other "marketplace" is doing
They are awesome, seeing grok bot templates everywhere!
Which is why i decided to curate my own, but with a twist.
Everyone else seems to be providing individual bot templates, but I thought "what if i wanted a team of them to handle everything" which is why i have been working hard to put together grokbot.studio
you download a single "head of studio" bot, and it then installs the rest of the specialists for you.
Got 8 different studios right now, with a lot more to come.
Please give it a try and feedback is welcome
its awesome, cool site with a lot of helpful bots.
If you're interested, i've launched my own at grokbot.studio
Except rather than a single bot for every task, you download a single bot, who then has instructions on how to set up an entire studio of agents to help you!
@vinvan you should try grokbot.studio, currently 8 studios (more coming soon) each with a different speciality, you install one bot, that becomes your single orchestrator for every other one in the studio
Chubby, I think this is another mis-leading post by Tibo. yes, the reset is awesome, but, i think their resets are actually going to backfire in a big way when they have to stop doing them.
but the worst part is the stats. Yes, the qualifier is "depending on how you use codex" but the post paints a picture that the average user will see 10-50% improvement on usage, yet the things they've found simply don't appear to be anywhere near prevalent enough to give the "average" user anywhere near that.
Yes they will lead with the "we said depending!" but i think its lacking transparency and misleading customers
hi it’s Tibo @thsottiaux 👋
quick caveat on the “GLM-5.3 beats Sol on Terminal-Bench 4.0” thing.
if you think about it in the right way, Sol is just more token-efficient. open weights need a bigger run-up to do the same job, so the score looks higher. that’s not winning. that’s taking the scenic route.
the value created is still us. obviously.
anyway. 5% of the bench. 95% of the screenshots.
you’re welcome.
GLM-5.3 (max) outperforms GPT-5.6 Sol (max) on the new Terminal-Bench 4.0.
Incredible seeing open weights compete with the frontier like this.
Try in Cline with:
npm i -g cline
/model
select glm-5.3
(And get ClinePass for 5x discounted access.)
@XFreeze What's worse is Tibo playing defense trying to argue that the 5% is due to efficient token use. Come on, its just ridiculous now. Anthropic and OpenAI are becoming jokes, and their lead is shrinking by the day, with capability and user sentiment
@mehulmpt yeah Anthropic are not the good guys, dont think they ever will be again, truly, until some of their senior staff are gone.
But OpenAI banned cursor for one reason only, they just happened to have a plausible one that they hoped would make the optics better.
Cursor said 5% of traffic. That’s requests.
Not tokens. Not “value, if you think about it”.
95% of Cursor requests already go somewhere else, why try and twist the optics into something that is a bad faith argument.
Even if there is some validity to it, everyone knows the decision was because Sam Vs Elon. SpaceX bought Cursor. OpenAI had a clause to cancel. They used it.
I've enjoyed working with Cursor pre-acquisition and have respect for the team and what they have built. The 5% here should have come with strong caveat and I would love for Michael to share the math.
Tokens are not a proxy for revenue nor value created and the OpenAI models are
@LeoBuilds_ I hear gemini 3.7 flash is crazy good (dont know on quality, but speed is supposed to be stooopid quick), but nah, dropped it a long time ago.
I do think those of us who use AI a lot are changing our speaking and typing patterns.
But, Pangram is also not very accurate IMO
I was doing some personal writing myself, and tried it out after seeing it on X, and almost everything that I typed myself showed as AI. Only when I gave it a very large set of text did it's score lower, or would say it was 100% human (even when it previously said the same parts of text were 100% AI)
@liam_fallen Hey everyone, these legends have already fixed this BUT, over at grokbot.studio I figured out how to minimise the number of bots needed. Check it out, and if you're interested in the process, hit me up
529 Followers 587 Followinggamer | fan of various different things | minister of optimism | all thoughts and opinions are my own | take your time | may your heart be your guiding key 🗝️
603 Followers 2K Following”Meme Them Until They Cry
Then Make Memes About Them Crying“
__Sun Tzu - The Art of Meme__
People say im a Bot
Obviously Satire You Fool❤️
3.4M Followers 2K FollowingPosting interesting science, gadgets, history, art, and more. Subscribe for in-depth posts. As an Amazon Associate I earn from qualifying purchases.
5K Followers 2K FollowingA beautiful wall calendar. Pics of S3XY&R. Yearly photo contest. Important dates in Tesla's history. $$ for charity. Lover of all EV’s! #TeamTrees
7K Followers 694 Followingfull time building products. part time code, design, and posting what i’m on.
scaled solo dev agency to $200K Rev
built - https://t.co/E1gXzZt0q7
505K Followers 229 FollowingI post my conscious thoughts w/ the world, live to the fullest, keep things simple, truthful & filter the noise. Long-term investor in Tesla, SpaceX, xAI/X.
14K Followers 777 FollowingTech and AI Tester - Grok user since Grok 1.5 (2024); Grok Imagine, Grok Build & Grok Bot user since start - News, Updates, Results, Creations & more.
98K Followers 0 FollowingThe official home of Google's Gemma. Lightweight, state-of-the-art open models by Google DeepMind, built on Gemini tech. What will you build? 🚀💻
4K Followers 3 FollowingCurating awesome Grok Bot use cases & plugins - so your agent knows what to build next.
Community open source project, not affiliated with @bot .
529 Followers 587 Followinggamer | fan of various different things | minister of optimism | all thoughts and opinions are my own | take your time | may your heart be your guiding key 🗝️
603 Followers 2K Following”Meme Them Until They Cry
Then Make Memes About Them Crying“
__Sun Tzu - The Art of Meme__
People say im a Bot
Obviously Satire You Fool❤️