JMT retweeted
We decoupled memory from the model.
Living Brain is one persistent brain for ChatGPT, Claude, Cursor, Codex and your own agents. Write in one, retrieve in any.
92.2% on LoCoMo. 464ms median retrieval. Up to 22.6x fewer tokens per question.
LLMs have parameters. Living Brain has plasticity.
JMT retweeted
GPT-6.2 Astra IS ANOTHER Level, it will be released today or tomorrow
I know a lot of people are going to be shocked by what this model is capable of Its intelligence understanding and overall capabilities are on another level
And that's not all GPT-6.2 Sol is also expected to launch alongside it together with another cyber model 馃憖
This will open a completely new era of AI We're talking about one of the most powerful internal models OpenAI has ever prepared to bring to the public I will be the first model to score above 60 on the benchmark
As I've been telling you, OpenAI has a LOT of surprises coming You're going to see it very soon
JMT retweeted
Binance will be "hacked"...
2026 or 2027.
There is no bottom in the market until Binance is hacked.
It's part of the script.
The hardest part of change is confronting who you are.
Without it, you can鈥檛 identify the gap to where you want to go.
1
226
it's stunning how much time and energy we're willing to spend doing everything except the work that would actually change our lives.
we keep ourselves busy, but avoid the uncomfortable changes necessary to create the life we say we want.
4
256
JMT retweeted
This is tweet of the month. If you find yourself ruminating repeatedly, it means your resources are not allocated correctly. You are wasting your brain's computation. Most of your time should be spent solving problems according to your goals.
If you are ruminating without working on a specific problem, you are like a badly written software program that chokes on memory. Most of your calculations are worthless.
Find ways to spend as much of your living time as possible solving defined problems with observable outcomes. If a problem is not well-defined, it should be. That will also cut rumination time (optimize your software).
basically everything in your personal life should be pre-computed and require ~0 thinking. if you're thinking lots daily outside of work it's more likely that it's a defense (intellectualization/rumination/rationalization) than actual thinking EXCEPT IF (see screenshot)
ALT https://meditationstuff.wordpress.com/2014/01/25/anti-akrasia-checklist-and-techniques/
You have a brilliant mind, an extraordinary ability to dream, to imagine possibilities, and to create ambitious goals. But your repeated inaction has slowly destroyed your self-trust, and with it, eroded your willingness to dream.
Not because you鈥檝e lost your imagination, but because you鈥檝e lost faith in your ability to follow through. You鈥檝e broken so many promises to yourself that dreaming no longer feels like possibility. It feels like another promise you can鈥檛 trust yourself to keep.
7
282
JMT retweeted
I've been running Haiku 5.5 non-stop for the last 12 hours, and this thing is impressing me even more than Opus 5.5.
First, no, i'm not saying that it is as good or better than Opus, its clearly not, but the results i am getting from this model, for a fraction of the cost, is astounding.
ChatGPT 6.1 Sol is left in the dirt by this model, never mind 6 Luna. Well done Anthropic and the Claude Devs
JMT retweeted
In the last week of September, DeepSeek models processed more tokens on OpenRouter than OpenAI, Google, Anthropic and xAI models combined.
