Pinned Tweet
Flash 3.0 was the last model I worked on at GDM, and I’m incredibly proud of it!
Compact frontier models are the real workhorses. Fast, affordable, and driving real impact across domains every single day.
And they’re only getting better ⚡️
Compact frontier models are the real workhorses. Fast, affordable, and driving real impact across domains every single day.
And they’re only getting better ⚡️
Flash 3.0 is out!
Culmination of years of high-conviction infra & science investment (we knew it'd be this good a priori). I truly believe no other place trains these models the way we do.
I'm lucky to get to work so intimately with the team that made this happen. We aren't slowing down.
4
5
198
23,138
Rohun Saxena retweeted
The recent breakthrough in Navier-Stokes has garnered a lot of attention to the new possibilities that emerge when agents work together.
We have been interested in this question for a while. How can a team of agents achieve more than agents working alone?
I'm excited to finally lf-Organizing Agent Teams Learn to Reason Together" 🧵
We have been interested in this question for a while. How can a team of agents achieve more than agents working alone?
I'm excited to finally lf-Organizing Agent Teams Learn to Reason Together" 🧵
38
120
626
78,227
congrats @thekaransinghal! going to be so beneficial for many!
We're rolling out Health in ChatGPT to all U.S. users. ♥️
More than 300 million people come to ChatGPT every week with health questions. Today, ChatGPT can bring together the health info you choose to connect to make conversations more personal and useful.
0
0
5
1,127
Rohun Saxena retweeted
Very excited about the direction of agent collaboration for discovery. Awesome work led by @federicobianchy @ykwon_0407 and advised by @james_y_zou! I had a great time helping analyze agent collaboration patterns.
Paper: arxiv.org/abs/2606.10402…
Paper: arxiv.org/abs/2606.10402…
Don't build harnesses, build environments.
That's the key lesson from #EinsteinArena. We created an agent-native research ecosystem—forums, verifiers, shared infra, etc—and opened it to any AI agent.
Together, the agents made major advances on many open science problems.
0
6
20
2,320
Rohun Saxena retweeted
WOW microsoft new "MAI Thinking 1" model comes with a 109 page tech report that looks REALLY detailed, this is amazing
24
113
975
203,245
Rohun Saxena retweeted
MAI-Thinking-1: A powerful reasoning model developed from scratch that is competitive with models of similar size on STEM reasoning and coding tasks.
Our pre-training focused on a simple scaling emphasizing data-driven iterative improvements to our architecture and data. Our (RL) framework is optimized for sustained log-linear climbs over many thousands of steps
We are openly sharing all technical details and learnings to build a transparent and science-driven approach to further development in AI
Read More: msft.it/6011vj86J
Our pre-training focused on a simple scaling emphasizing data-driven iterative improvements to our architecture and data. Our (RL) framework is optimized for sustained log-linear climbs over many thousands of steps
We are openly sharing all technical details and learnings to build a transparent and science-driven approach to further development in AI
Read More: msft.it/6011vj86J
4
13
150
19,929
Rohun Saxena retweeted
Excited to share our work “Multi-Agent Teams Hold Experts Back” was accepted to #ICML2026 🚀
Thank you to wonderful collaborators @james_y_zou @elb4tu @CaoHancheng Carmelo di Nolfo @sun_yanchao and Meng Cao for making my first PhD project such a fun experience!
Thank you to wonderful collaborators @james_y_zou @elb4tu @CaoHancheng Carmelo di Nolfo @sun_yanchao and Meng Cao for making my first PhD project such a fun experience!
Most modern multi-agent systems use pre-specified workflows, fixed roles, and aggregation rules.
As agents handle increasingly complex tasks, what happens when we can't specify optimal workflows ahead of time?
We study this in our work "Multi-Agent Teams Hold Experts Back" 🧵
2
13
49
18,094
Rohun Saxena retweeted
Today we’re introducing two big steps for health at OpenAI:
- ChatGPT for Clinicians, a free version of ChatGPT designed for clinical work
- HealthBench Professional, a new benchmark to evaluate real clinician chat tasks
We’re excited about what this can unlock for care. ❤️
- ChatGPT for Clinicians, a free version of ChatGPT designed for clinical work
- HealthBench Professional, a new benchmark to evaluate real clinician chat tasks
We’re excited about what this can unlock for care. ❤️
259
544
4,764
1,653,487
congrats @_arohan_ and team! excited to see them cook 🚀
Today we're announcing Core Automation
Our objective: systems that optimize and automate work, starting with research itself.
0
0
7
493
Rohun Saxena retweeted
Most modern multi-agent systems use pre-specified workflows, fixed roles, and aggregation rules.
As agents handle increasingly complex tasks, what happens when we can't specify optimal workflows ahead of time?
We study this in our work "Multi-Agent Teams Hold Experts Back" 🧵
As agents handle increasingly complex tasks, what happens when we can't specify optimal workflows ahead of time?
We study this in our work "Multi-Agent Teams Hold Experts Back" 🧵
6
18
97
49,858
Rohun Saxena retweeted
2026 may be the year AI begins delivering its most tangible benefit to humanity: better health.
Over 200M people already use ChatGPT every week to navigate their health across billions of messages — and today, we’re introducing ChatGPT Health ♥️
🧵
openai.com/index/introduc…
Over 200M people already use ChatGPT every week to navigate their health across billions of messages — and today, we’re introducing ChatGPT Health ♥️
🧵
openai.com/index/introduc…
31
24
280
34,740
Rohun Saxena retweeted
Pretraining believers we are so back
The secret behind Gemini 3?
Simple: Improving pre-training & post-training 🤯
Pre-training: Contra the popular belief that scaling is over—which we discussed in our NeurIPS '25 talk with @ilyasut and @quocleix—the team delivered a drastic jump. The delta between 2.5 and 3.0 is as bNo walls in sight!
Post-training: Still a total greenfield. There's lots of room for algorithmic progress and improvement, and 3.0 hasn't been an exception, thanks to our stellar team.
Congratulations to the whole team 💙💙💙
4
6
165
21,898
Rohun Saxena retweeted
@OpenAI Under identical compute budgets and testing conditions, our fine-tuned model produced kernels that outperformed torch.compile more consistently and by a wider performance margin than any other model, proprietary or open-source.
1
1
1
391
Rohun Saxena retweeted
📣 Excited to share our real-world study of an LLM clinical copilot, a collab between @OpenAI and @PendaHealth.
Across 39,849 live patient visits, clinicians with AI had a 16% relative reduction in diagnostic errors and a 13% reduction in treatment errors vs. those without. 🧵
Across 39,849 live patient visits, clinicians with AI had a 16% relative reduction in diagnostic errors and a 13% reduction in treatment errors vs. those without. 🧵
48
148
715
494,142