exploring ai and agents, testing every new model, and writing for ai brands
inyourmind
Joined July 2022
- Tweets63.4K
- Following661
- Followers2.4K
- Likes118K
Be Anthropic
- give Grok Bot access to Opus 5.5 and let it study how Grok Bot works, learn its strategies, and adapt to everything that makes it effective
- Then build your own bot using same Opus 5.5
Congratulations, you are now Anthropic
O🌷 retweeted
$200 in USDG up for grabs. 👀
20 winners. $10 each.
Paid privately through gloamtrade.
Here's the interesting part:
Your onchain payments are public by default but Gloam just brought real privacy private payments to two chains today.
To enter:
→ Follow @gloamtrade
→ Like and repost this post
→ Reply with one reason your onchain payments should be private.
Winners will be selected in 12 hours and contacted via DM.
Claude might have just killed the idea of reverse engineering anything with AI
Someone tried to get Claude to research REA, summarize everything, and turn it into a detailed guide.
Claude completed the research, then triggered a safety filter (safeguard that wasn’t there before) halfway through writing and refused to finish, even after rewording!
We really need Open-source to take over!!
Fable 5.5 is coming today (block everyone saying this)
they don’t know what they are saying
Fable 5.5 is coming next Tuesday
Mark my words
Gemini 4 Carbon did this!! (NOT Argon)
It seems Arena is currently routing Gemini 4 Argon to Carbon, so you might have access to it there
You might want to test it yourself and see if you notice any difference between the two
Personally, I think Carbon is noticeably better at 3D generation and shading
should i drop Argon vs Carbon?
No Fable 5.5 bluds
Go Home
New on our Science Blog: designing protein binders against scorpion neurotoxins.
Scorpions sting about 1.2m people every year, and antivenoms are often costly or in short supply. Claude has made an early step toward finding new treatments.
anthropic.com/research/claud…
I can’t be the only one who noticed ChatGPT getting worse day by day
It can’t even solve the simplest questions in a single response anymore, it keeps going in circles, repeating the same things, and giving me disappointing results
Things it used to solve without any problems are suddenly becoming a struggle, the usage limits, the performance, the quality of responses, everything feels worse
I’m using GPT-6 and it feels like I’m using GPT-3.5 lmao
it’s getting frustrating tbh
holy, someone built a game where you have to identify real players hiding among NPCs and it looks insane
it is a multiplayer game where some characters are actual NPCs and others are players pretending to be NPCs
as a player, you have to blend into the crowd, copy NPC movements, complete objectives, and avoid getting caught
as a hunter, you have to observe everyone and figure out who is secretly a real player
imagine playing this with your friends lmao
O🌷 retweeted
Soon you won't need to visit Google or Wikipedia.
Your agent will do it for you and tell you what it found.
🤝 Paid partnership
We raised $9M to give your agents the internet.
Led by @generalcatalyst, with @orangecollectv and Robinhood Ventures.
Search, scrape, crawl, monitor with one API trusted by 800+ customers.
We’re making it ridiculously easy to build with web data.
Opus 5.5 just recreated super mario 64 for the game boy advance
it has 4 playable levels, 3d graphics rendered entirely in software, and mario’s original movement and collision physics
also it runs at 30 fps on 98% of frames in the mgba emulator
repo: github.com/Kodoque1/sm64-…
deepseek v4.1 flash-1017 just built an interactive moon base you can explore in your browser (from a single prompt)
it generated the entire thing using three.js in one html file, with zero manual edits
the crazy part is the speed
it was running at nearly 200 tokens per second and the entire generation cost only a few cents
all from one prompt at max reasoning effort btw
source: github.com/tetsuo-ai/agenc-c…
