Nishit Patel (@nishit130) header
Nishit Patel (@nishit130) profile picture

Founding Engineer https://nitter.cf/t.co/lD9ppNa5ZL | GSoC 21 @GNOME

Joined February 2017
asked claude code a random question about embeddings. before answering it checked secondbrain and pulled up a paper i'd saved weeks ago that answers it exactly. i'd forgotten i had it. this is the whole point -> my ai reaches for what i've read, on its own.
2
11
267
After watching all the Adipurush memes, I feel like watching it just for the laughs. Ig it would be hilarious 3 hours 🤣🤣
45
Building Contextual Recall into secondbrain.sh. As you read, a margin dot appears beside every passage that overlaps with something you've saved. Hover to see what you saved and when. Your old reading, back exactly when it's useful. Join the waitlist → secondbrain.sh
1
2
81
🏡 Room for a male flatmate in 4 BHK duplex, BTM 2nd Stage, from Nov 1 
₹20K rent | ₹50K deposit 
Cafes nearby, green street, rooftop sunsets 🌇 
Close to HSR, JP Nagar, Koramangala
DM me! @BangaloreRoomi @flatshareblr @GruhamBot @MIBangalore @fmrbangalore
6
2
1
9
1,789
still early, waitlist's open 👉 secondbrain.sh
2
30
I keep losing thoughts 🧠 notes, screenshots, random docs, half-written tweets… building secondbrain.sh so everything i know is one search away. Still early. waitlist is open if you want in👇 secondbrain.sh
8
2
16
657
Feel like this post will age like fine wine.
I resigned from Anthropic today. I spent the last three years doing pretraining research at both OpenAI and Anthropic. Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives. More thoughts below.
1
90
Benchmark numbers for GPT-6 Astra look really insane..
1
2
122
When leaders or political parties believe they are untouchable, it's time for a change in power. The real issue now isn't just a debate of right and wrong, but the blatant lack of empathy from the top, acting like everything is fine while ignoring harsh realities.
6
174
The only thing more dangerous than a person who knows everything is a person who thinks they know everything. Stay curious, stay humble. 🧠✨
5
10
381
Must-read paper!
we distilled 2.3M Claude Fable 5 reasoning traces into Qwen3-4B - 100% self-consistency @ 512 samples - 0.00 bits output entropy - zero hallucination variance turns out the student is not bounded by the teacher. it also converged on one universal truth. we open-sourced the model weights👇
1
1
5
1,701
interesting!
Announcing a new division of Midjourney called "Midjourney Medical"
32
Small, specialized models always outperform the big ones. They are trained for very specific tasks using high-quality data. In this case, verifiable reasoning, such as math and coding. However, when it comes to overall general intelligence, they still won't match the big mammoths.
WHAT THE HELL is happening in AI? A 3B parameter model just put up coding benchmark scores in the same league as Claude Opus 4.5. 3 BILLION. The weights are on Hugging Face, anyone can test it. I genuinely don't know if this is a breakthrough or if the benchmarks are broken.
3
54
So much hype, was anyone able to do anything meaningful with fable without running out of limits?
The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance. Access to all other Claude models is not affected. We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible. Read our full statement: anthropic.com/news/fable-myt…
1
59
Google ended I/O 2026 by saying we’re standing in the "foothills of the singularity." 🏔️ They aren't talking about "helpful assistants" anymore. They’re talking about replacing the human decision making loop entirely. We’re not building tools anymore, we’re training our own replacements. The "Agentic Era" is just a fancy term for large scale agency outsourcing. #ai #google
2
70
12 Million tokens. 52x faster than FlashAttention. <5% the cost of Opus. If SubQ’s new sub-quadratic architecture is the real deal, the context window wars just broke the sound barrier. 🔥👇 @subquadratic
Replying to @alex_whedon
We were a little slow on this, but we just got a technical blog post up with more details. Please take a look! subq.ai/how-ssa-makes-long-c… We have a model card coming next week, and we are happy to take requests for any specific details there. I am happy to answer any questions here!
4
158
Unpopular opinion: Dedicated vector databases are the biggest grift of the current AI boom. Stop deploying massive, heavy infrastructure just to do basic RAG. PostgreSQL + pgvector does the job for 99% of you. You are not building Google.
1
3
73
Whatever Trump says and does is the classic case of "Brinkmanship strategy". Right now, a Nash equilibrium of Mutually Assured Destruction (MAD) has been achieved. The only way to break nash equilibirium is by abandoning it. Brinkmanship is the strategy of deliberately pushing the situation closer and closer to the edge of disaster to force the opponent to back down. In game theory terms, you try to convince the other player that you are irrational. If Iran belives trump is crazy enough to press that button, it might surrender before the button is pressed, breaking the Nash equilibrium. Do you think Iran will fall for it?
🚨 BREAKING: Trump warns 'whole civilization will die tonight, never to be brought back again' if Iran doesn't agree to deal to end war
5
99