If you've connected your Grok Bot to Whatsapp, send this prompt:
"Given what you've learned having access to my Whatsapp, what would you estimate my family members to be?". I am 100% aware I shared a lot of data...but it's still mind blowing to read the results.
8
It is so frustrating that out-of-context soundbites get more traction than the actual truth from the source.
but only because...i actually do want the world to have tons of memory and people to put as much as they want in their computers, and it is fucking annoying that this was taken out of context.
i have been arguing for the world to ramp up memory production for a long time.
22
Muse Wonder Full is really weird to use in Portuguese. It doesn't sound natural at all and seems to be closer to Portugal Portuguese instead of BR which is where I'm at and the one I use. Claude, GPT and Gemini don't have this issue.
1
51
GPT 6.1 Terra XHigh on Fast is becoming my main model...if I need code or more complex work I have it delegate to Sol or Astra...but for *a lot* of things I can just work closely with it, given how fast it is and get a lot done.
1
1
85
Opus 5.5 and Sonnet 5.5 are INSANE, Astra and Sol 6.1, and Video Editing... youtu.be/VQffPizYPL4?is=oHss… via @YouTube
Outstanding show, real builders here.
32
@thsottiaux if the rumors on Terra are true, you should definitely get the dots to look like super heroes and while you can’t directly use avengers, you can just say Terra 6.1 is the avengers version along with allowing multiple dots. This would be hilarious :)
19
Maybe it’s a skill issue and I don’t know how to use Gemini 4? This demo is INSANE
Argon has this amazing natural understanding of the world. It can do so many things from first principles.
Here's a pelican riding a bike, using fourier epicycles.
Video made by Argon.
Time-synced Lyria 3.5 music prompted by Argon.
42
The unfortunate summary of testing Gemini 4 Argon:
"I found the session (“Adversarial Code Review and Testing”). The Opus part was solid. The Gemini part said “all fixes done, only a 1-line test fix left”, but that wasn’t true, so I picked it up from there."
@OfficialLoganK Can we get /feedback on antigravity? I would gladly share the session and all that has happened there as feedback, I would really like to see Gemini continue to improve.
2
73
Muse Wonder...I wonder what the output would be. I don't have an expensive sub but I have not touched mine since Opus 5.5 came out...my first try of just getting feedback on a new projetct...I wonder what the output would be. @finkd
"You are logged in.
Contributor mode is on, so your content may be used for product improvement. Turn it off in /settings. See Terms of Service for more information.
Model set to muse-wonder-1.0
❯ I would like you to give me an indepth review of this codebase.
◆ Read skill workflow-authoring · built-in · 32 KB
◆ Workflow(In-depth codebase review) · ctrl+o
└ Running in background · /workflows to monitor
(↓ to view workflows)
◆ Finished In-depth codebase review — completed · 0/10 completed · ctrl+o
◆ model failed: API error 429 [request_id=89b1682e-cfb0-4e37-a1b8-4b392db8ed33]: Subscription quota exhausted. Your usage window resets at 2026-10-13T05:28:05Z. (rate_limit_error)"
44
When Claude models start doing this: "Honestly: two of the three I should have fixed rather than listed, and the third needs a decision from you. I parked them because I was closing out the run, which isn't a good enough reason for the first two." you just know something is off
1
50
Just had my first @OpenAI dots magical experience. I’m really bad with my personal email as it’s mainly a collection of Spam. But now that I have dots looking at it, I’m getting actually helpful things out of it. I was tagged on a GitHub thread from a team member I would have missed and not only it found it but had context to help me deal with it, as I messaged TARS (my dot) back it just took action on gh. I went from potentially missing something to addressing it with the simplicity of a message.
72
SpaceXAi launched Grok Bot, a lot of attention and focus on it. Grok 4.7 wasn’t the improvement we expected (it’s a decent model, I use it)
OpenAI launches Dots and gives people unlimited Astra and now the *fast* tier for GPT 6.1 is 30 tokens per second.
I appreciate how they are pushing and creating new ways of doing things…but the price on doing all in parallel is clear…
Anthropic currently has Opus and Sonnet performing incredibly well. No focus lost to “pets” and what not.
As customers of it all, we win. We have choice and optionality. At the moment, Anthropic delivers what I need and helps me get things done and I prefer it over the other solutions.
1
60
There is a big difference between the AI Users, those that use it to do things and those that are building systems using AI. Their perception of which model is good or bad, cost, speed, etc…is wildly different. Know who you are following so you can attribute more or less weight to their feedback and experiences.
1
20
Grok 4.7 feels like an unpolished version of a good model. It's generally better than 4.6 but it has some quirks, if you give it too many commands at once it can get confused, etc...Grok 4.6 was a super polished version of a smaller model so it felt that way. I hope they now go back to optimizing for token efficiency as with this speed and less tokens this model will be an absolute no brainer for daily work.
1
1
62
@XiaomiMiMo Ultraspeed was the release of the day IMHO. Really looking forward to using it. Is there a status page to see how the servers are doing? I was trying to use it, and while it eventually processes quickly, it's taking minutes to start.
1
31
