Professor scientist, immunologist, biomedical engineer & Biohacker. ALL IN ON AI & ROBOTICS! #Longevity #BioAI #Codex #Space #Scifi #Singularity #trekkie 🖖

Connecticut, USA
I am incredibly honored and humbled to be featured in @OpenAI’s new Codex video ☺️🙏
61
35
725
23,364
I asked Claude Opus 5.5 to make a 2-minute video on the history of immunology. Every frame & every note of the music was generated in code. It's really good! As an immunologist, this made me emotional. So many lives saved by vaccines, built on research we often take for granted
4
4
50
2,364
Derya Unutmaz, MD retweeted
AI painting of the day. Created with ChatGPT image 2.5
Made with AI
9
6
122
10,694
Derya Unutmaz, MD retweeted
I can confirm that Dot is amazing! The more I use it, the more I like it. For me, the key impact is that it’s much more intelligent than any other AI agent I've tried. Also, as Sam points out, it learns about you & gets more proactive. I'll soon share some really cool use cases!
dot is my favorite openai product so far! it is amazing to me that each day it feels noticably better as it learns more of my workflow and style. having it do the stuff i don't like doing--and usually just builds up as a gravity well of dread--has me very happy.
29
12
253
16,873
This is an incredible ChemAI benchmark milestone! GPT-6.1 Sol is now at the frontier of AI benchmarks for drug development. This is remarkable, especially because it is not a general model and was not particularly or specifically trained for multi-step retrosynthesis!
☀️ Breaking! We have new SOTA model in our URSA OOD benchmark for multi-step retrosynthesis! And it is GPT-6.1 Sol, medium with 35% solved molecules. Just less than a week ago we got result from GPT-6 Sol that recapitulated RetroChimera level (32%). Amazing! @sama @gdb 👏 1/2
7
6
130
11,024
Derya Unutmaz, MD retweeted
I’m sharing a video of the new version of the DOOM-like game I've been making with GPT-6.1 Sol! Warning: This one is more gory and much bloodier, just like the original DOOM 😅! It also plays almost exactly how I remember it from 30 years ago! It’s a Mac game, but I’ll be sharing it on GitHub soon!
15
8
122
11,884
So cool! Dani keeps cooking, for real and with dots! ☺️ It was great seeing y’all at OpenAI DevDay!
while our dots worked, @axbehr and I literally COOKED with @MatthewBerman and @ForwardEditor 👩🏻‍🍳
4
2
118
10,845
Great to see @DeryaTR_ at #ARDD2026, where he's speaking today on "How AI can help defeat aging: Roadmaps, breakthroughs and what's next." Looking forward to it! @ARDD_meeting
2
3
21
3,186
Derya Unutmaz, MD retweeted
Sarah created this amazing comic book for us based on the X-Men series, one of my favorites. I was the regenerator, exactly the superpower I'd want! Super cool. Thank you, Sarah!
Such a fun DevDay at @OpenAI So inspiring. 🙏 @DeryaTR_ @craigsdennis and so many others
2
4
36
11,437
Derya Unutmaz, MD retweeted
Wow! Wow! @GoogleDeepMind is back with Gemini 4 Argon! Insane benchmarks! In one fell swoop, they’ve jumped to the very top of the AI frontier. As I’ve said before, never bet against Google in the age of AI! Can’t wait to try Gemini 4 ASAP!
Introducing Gemini 4 Argon – our new frontier model. It’s built for complex workflows across coding, enterprise knowledge work, and cybersecurity defense – rolling out today to a set of trusted testers through our Fairwind Program.
18
36
422
24,557
Derya Unutmaz, MD retweeted
AI painting of the day. Created with ChatGPT Image 2.5
9
6
118
12,157
Derya Unutmaz, MD retweeted
WOW! Gemini 4 Argon just reached #1 on Blueprint-Bench 2, a benchmark that measures AI’s understanding of the physical 3D world. It is now almost at human-level performance, which is absolutely astonishing considering that just a year ago, AI models were scoring close to zero! In fact, I believe this is one of the most critically important benchmarks for advancing to superintelligence. Understanding the physical 3D world is foundational to interacting with it, making spatial intelligence essential for both robotics AND biology! Most people probably don’t immediately recognize the implications for biology. Yet biological systems operate in 3D space and evolve over time, making them fundamentally 4D, with temporal dimension. Everything in biology depends on spatial organization and dynamic interactions: proteins folding and interacting, cells communicating, immune cells navigating complex tissues, and entire organs functioning as highly coordinated, dynamic systems. Imagine AI models with advanced spatial intelligence combined with a deep understanding of biological mechanisms. Such systems could dramatically accelerate the development of virtual cells, patient digital twins, autonomous scientific laboratories, and ultimately, simulations of entire human organisms. This brings us closer to the ultimate dream: curing all diseases and reversing aging! What makes this so exciting is that AI is advancing toward superintelligence with the potential to understand, model, simulate, and eventually manipulate both the physical and biological worlds. This may be one of the final bosses on the road to scientific superintelligence: AI that understands the world deeply enough to make groundbreaking scientific discoveries, design entirely new biological systems, and transform our ability to engineer life itself!
Is Google so back? Their new Gemini 4 Argon understands the physical 3D world better than any other AI. It ranks #1 on Blueprint-Bench 2, a benchmark where AI agents draw floorplans from photographs of apartment interiors.
13
22
318
18,103
Derya Unutmaz, MD retweeted
Back to solving aging with my dear friend @biogerontology & his company @InSilicoMeds, who organized the largest aging/longevity meeting I've ever seen, ARDD, at @Harvard! This is now ground zero, where we begin the revolution to solve & reverse aging with superintelligence!
13
22
246
14,279
The video Turing test has been passed with this Tavus AI model. AI agents will now have realistic-looking personas, and most people won’t be able to tell them apart from humans. The impact will be enormous, to say the least!
Introducing Griffin, the first model to pass the video Turing test. 48% of people who talked to it live thought it was a real human. Previous systems have had a pass rate <3%. It is #1 on NVIDIA's benchmark for full-duplex AI video. It’s the first Human Interaction Model (HIM).
Community note
The 48% figure and "video Turing test" claim are from Tavus's own study of 54 one-minute calls, not independently verified or using a standard protocol. Griffin-Lite leads NVIDIA's VideoFDB benchmark on their public leaderboard. cellcog.ai/blog/tavus-gri… research.nvidia.com/labs/amri/proj… tech-ish.com/2026/10/02/tav…
13
8
171
22,678
Derya Unutmaz, MD retweeted
I’m making this cool aquarium game with GPT-6.1 Sol & it’s already pretty fun to play! You can feed the fish, maintain water quality, change habitats & add new fish. I’m now creating smaller breeding tanks for different fish species, each with biologically authentic breeding behaviors and environmental requirements!
9
9
158
12,551