Sophont
@SophontAI

Open medical superintelligence. WE'RE HIRING! (see website)

sophont.med
Joined March 2025
38 Photos and videos
Pinned Tweet
Sophont
@SophontAI
Sep 21
This year, we released Medmarks v1.0, our open benchmark suite + leaderboard of LLM medical capabilities. Today, we release additional results for some of the latest mid-size open-source LLMs including Gemma, Qwen, Muse, Nemotron. We find Gemma 4 31B leads in this size class. Link: sophont.med/blog/medmarks-su…
3
3
17
7,366
I'm excited to share Medmarks was accepted to the NeurIPS datasets and benchmarks track! 🥳 See you at Sydney!
Sophont
@SophontAI
May 12
We're excited to release Medmarks v1.0 + a technical report! This is an update to our Medmarks benchmark suite, the largest open-source automated suite for evaluating the medical capabilities of LLMs. We added 10 benchmarks (20→30) and 15 models (46→61) to the leaderboard!
9
8
77
5,365
NEW RESULTS: Our researcher @benjamin_warner has been diligently working on updating the Medmarks leaderboard (evaluating model's medical capabilities) with some of the latest mid-size open-source models. Some interesting results, read to find out more:
Sophont
@SophontAI
Sep 21
This year, we released Medmarks v1.0, our open benchmark suite + leaderboard of LLM medical capabilities. Today, we release additional results for some of the latest mid-size open-source LLMs including Gemma, Qwen, Muse, Nemotron. We find Gemma 4 31B leads in this size class. Link: sophont.med/blog/medmarks-su…
1
5
23
4,499
We are in the process of evaluating more models for the next release of Medmarks, including Astra and Fable, but first, a teaser.
Sophont
@SophontAI
Sep 21
This year, we released Medmarks v1.0, our open benchmark suite + leaderboard of LLM medical capabilities. Today, we release additional results for some of the latest mid-size open-source LLMs including Gemma, Qwen, Muse, Nemotron. We find Gemma 4 31B leads in this size class. Link: sophont.med/blog/medmarks-su…
2
2
9
1,272
This year, we released Medmarks v1.0, our open benchmark suite + leaderboard of LLM medical capabilities. Today, we release additional results for some of the latest mid-size open-source LLMs including Gemma, Qwen, Muse, Nemotron. We find Gemma 4 31B leads in this size class. Link: sophont.med/blog/medmarks-su…
3
3
17
7,366
more replies
Finally we analyze the Nemotron series of models. We see steady progress across generations.
1
1
101
We hope these results on these open-source models are useful to the medical AI community. We're finishing up evals on the recent frontier and near-frontier models, including planned evaluations of Claude Fable 5.1 and GPT 6 Astra. Look for the full update in the coming weeks.
2
100
.@SophontAI fixes this :)
Tanishq Mathew Abraham, Ph.D.
@iScienceLuvr
Sep 2
notice how everyone in the tech community likes to say AI will cure cancer but almost none of them are actually working on it...
1
1
16
3,617
The Secret Sophont Master Plan (just between you and me, and in a single tweet)
Tanishq Mathew Abraham, Ph.D.
@iScienceLuvr
Aug 14
At @SophontAI, the plan is simple: 1. build encoders for every single modality in medicine and biology 2. align the latent spaces of all the encoders into a unified patient representation 3. use the unified patient representation to diagnose and treat patients 3. PROFIT
1
4
1,915
Forgot to share, but @SophontAI recently received an Honorable Mention for the @nebiusai AI Discovery Awards 2026 :)
6
1
38
3,516
Sophont retweeted
Jeff Dean
@JeffDean
Aug 8
Replying to @iScienceLuvr @SophontAI
You're all doing amazing stuff! Keep it up! I'm delighted to be along for the ride.
6
4
266
25,112
Btw a reminder that @SophontAI is actively hiring :) sophont.med/hiring
11
18
144
14,233
Btw since there were a bunch of Jeff Dean stories being shared recently, I thought I'd share mine: Jeff Dean was the second investor in @SophontAI. We got connected through a founder friend. After briefly reviewing our pitch deck, he immediately decided to invest. He had a high level of conviction in our mission much earlier than many others. Something I was honestly surprised about is his level of support for our company. As the Chief Scientist at Google, I figured he would be extremely busy but whenever we have any asks of our investors he quickly responds with suggestions and he always tries to find ways to help. I don't know how he has time to do it! I am so grateful and honored to have Jeff Dean as an investor in Sophont, not just because he has been a massive inspiration for me as an engineer and medical AI researcher, but because of his early conviction and steadfast support of our mission.
30
38
1,111
134,259
We had a great ICML 2026 in Seoul! * We presented CortexMAE, our paper on fMRI foundation models and benchmarking * in the Foundation Models for Life Sciences workshop we presented Medmarks, our paper on LLM medical benchmarking * We had a widely successful lunch social cohosted with Ricursive Intelligence, GMI Cloud, 1943, Z Potentials Appreciate all the interest and thoughtful discussion about our research, see you at the next conference!
2
5
16
3,691
We had an incredibly successful and popular ICML lunch social last week in Seoul, with lots of great conversations and food! Thanks to everyone to showed up. And thanks to our amazing cohosts!
Lillian Ma
@lillian_ma_
Jul 15
Has everyone made it home safely from ICML? ❤️ Thanks to everyone who joined our Bits & Atoms lunch. The cohost group photo turned out way too cute, and it feels like the perfect ending to an amazing ICML. Nothing can replace meeting friends IRL. I’ll remember every face I had the chance to talk to, and I can’t wait to see you all again at NeurIPS in Sydney. @gmi_cloud @SophontAI @RicursiveAI
4
2
34
5,714
Come check out our ICML workshop poster on LLM medical evals RIGHT NOW
Tanishq Mathew Abraham, Ph.D.
@iScienceLuvr
Jul 5
Visiting Korea for the first time for ICML 2026! DM me if you want to chat about medical AI. self-supervised learning, post-training, etc. Also a reminder we are hiring :) We also have a main paper and workshop paper, please check them out! 1. "Scaling Vision Transformers for Functional MRI with Flat Maps" - Wed 10:30am 2. "Medmarks: An Open-Source LLM Benchmark Suite for Medical Tasks" - FM4LS workshop, Sat 11:05am
2
5
42
6,949
We are presenting Medmarks at the Large Language Models for Life Sciences workshop at ICML today at 11:05am poster session. Come on by!!
Sophont
@SophontAI
May 12
We're excited to release Medmarks v1.0 + a technical report! This is an update to our Medmarks benchmark suite, the largest open-source automated suite for evaluating the medical capabilities of LLMs. We added 10 benchmarks (20→30) and 15 models (46→61) to the leaderboard!
2
14
3,728
Reminder that we are cohosting a lunch social at ICML today! Please join!! (Link in reply)
1
2
8
4,228
We are presenting this work RIGHT NOW till 4:15pm at ICML poster session #806
Sophont
@SophontAI
May 28
NEW RELEASE: Today we're releasing CortexMAE: a family of fMRI foundation models trained on 2.1K hours of open fMRI data. We're also releasing Brainmarks: an open benchmark suite for evaluating fMRI foundation models. Full paper is on arXiv (accepted to ICML 2026) A thread:
1
6
19
3,993
Load more