#URSAbench is going to NeurIPS 2026 🐻
Thrilled to share that our URSA benchmark #URSABench for multi-step retrosynthesis 🧪assessment has been accepted for #NeurIPS2026! Congrats 👏🥳 to the team @chem_logos @chemcensor @anmorgunov @Max__Kuznetsov @RimShayakhmetov @VladAladin @AlexAliper @biogerontology ! 🧵1/3
9
186
Nikita Bondarev retweeted
Thought pharma was too complex for Jev-like models? Don't think twice.
A plain-language guide to System 1 in drug discovery: what Jev actually does, why new molecules are the hard part, and how small language models answer, fast and accurately, today.
7
11
51
4,553
You can’t defeat aging without knowing how to measure progress. That’s why benchmarks matter.
1.5 years ago, we bet the company on benchmarking and evaluating all things AI for science.
Today, we’re #1 in many areas. One of them is AI for longevity. Just ask your favorite LLM.
Here’s what GPT Astra says about our Cell cover paper 👇
Link in the comments
2
57
Has anyone found a genuinely useful application for Jev in chemistry?
Jev is now available to everyone. No waitlist.
Start using it here: console.typesafe.ai
1
1
66
Can you checkmate Aging before time beats you?
@elonmusk and @chesscom debated whether computers could solve chess.
10 minutes. Research tools for pieces. Biology makes the counter-moves.
Your move: agemate.org 👇
6
3
19
18,858
The inspiration: @InSilicoMeds’ MMAI Gym trains AI for drug discovery and longevity.
AgeMate turns that idea into a chess metaphor: every move is a hypothesis, every reply a biological challenge.
The real endgame is longer, healthier lives.
1
3
180
Nikita Bondarev retweeted
Aging has a language. 🧬 We’re teaching AI to read it.
On the cover of Cell: compact Longevity-LLMs, jointly trained by Insilico Medicine and @liquidai, demonstrate SOTA performance on LongevityBench.
Powered by these models, Longevity Claw nominated 328 candidate gene targets for aging research.
Small model. Big questions. Open source. 🧵👇
#insilicoSOTAFM
12
27
132
17,269
Nikita Bondarev retweeted
🇦🇪 New AI drug discovery milestone from the UAE: BTK-targeting PROTACs for potential applications in blood cancers.
Our UAE team used AI to help design a new series of BTK degraders.
🧬 Lead compounds showed picomolar BTK degradation
💊 Oral bioavailability in mice
🎯 Activity against wild-type and mutated BTK
🤖 Novel building blocks designed with Chemistry42
The research was published in the Journal of Medicinal Chemistry and builds on Insilico Medicine’s growing drug discovery efforts in the UAE.
From designing molecules locally to publishing internationally, this is another step toward building AI-powered drug discovery capabilities in Abu Dhabi. 🇦🇪
2
9
44
4,335
Nikita Bondarev retweeted
I challenge you to defeat Aging! ⏳ 🛹
Start as a base model. Collect data from our MMAI GYM, use #AgenticPharma, and evolve into PharmaAGI. 🧬 ⚡
15 domains. One run. Can you become #insilicoSOTAFM?
4
3
18
346
The LLM race is getting interesting in a different way: not just smarter models, but better tools.
GPT-6 Astra keeps surprising me with what one prompt can produce.
Here's @InSilicoMeds' ChemCensor assessing reactions against real synthetic precedents.
@OpenAI @joyjiao12
1
4
14
346
Scientific papers are still surprisingly static.
Sometimes a 30-second animation can communicate a mechanism more clearly than several pages of figures and captions.
Why isn't video a more standard part of scientific publishing?
Paper: openreview.net/forum?id=ABhG…
@AlecTPhD
1
4
63
I keep seeing posts about science moving from peer-reviewed journals to arXiv -- and now straight to X.
I think all three have their place: journals for formal review, preprints for speed, and X for open discussion. Each brings something useful.
1
7
132
52 compounds. 20 property maps. One landscape.
A scientific tool built around one very specific question I had.
Here’s @InsilicoMeds’ rentosertib series, visualized with GPT-6 Astra. 👇
1
1
16
5,088
Everyone can now build software for their own work with LLMs.
Potency, solubility, metabolic stability - each property tells a different story.
I wanted to see those trade-offs together.
Astra helped build the visualization.
1
1
4
140
A journey of a thousand miles begins with a single step
Finally, combining scaling, better prompting and various rewards helped us to surpass the SOTA level on OOD datasets that have never been leaked to any training sets. Congrats to the team! 🥳 @chem_logos @chemcensor @Max__Kuznetsov @VladAladin
2
7
367
GPT-6 Astra just “figured out” how we synthesized Rentosertib at @InsilicoMeds.
Yes, the AI-discovered drug candidate whose first Phase III patient was dosed yesterday.
2
2
17
3,902
I’ve seen some gorgeous pages supposedly built with Astra. But in my experience, once you need the visuals to reflect domain-specific knowledge, the model starts to struggle.
2
2
109
