Advay Pal retweeted
Today we’re excited to introduce Surgeon, our first model trained to assess how surgical a contract redline is.
The best lawyers can change the entire meaning of a document with just a few words. It’s an art form. We’re teaching models to recognize it.
It's already live in production for our customers.
as someone who grew up with a bar in the living room, a gym in the living room is even better for hosting people
a journalist, 3 physicists, an engineer, an english lit major and a boy who didn’t go into college walk into a bar…
many years ago @Caltech physics professor Tom Tombrello told @elonmusk about a student of his, “the only one in my 50 years who reminds me of Feynman”
that student was @DarioAmodei
Advay Pal retweeted
In just two weeks, @altaratech helped a semiconductor materials provider increase yield from 80% to 94% on a capacity-constrained production line.
To meet surging semiconductor demand, Polar Performance Materials needed to rapidly scale production of its ultra-high-purity alumina without compromising quality.
By combining more than a decade of process knowledge with Altara’s AI platform, Polar was able to translate its accumulated materials and manufacturing expertise into higher yield, greater productive capacity, faster problem solving, and substantially higher responsiveness to customer needs. Key outcomes included:
- 14-percentage-point yield increase in 2 weeks
- 1,000+ hours saved across R&D, manufacturing, quality, and engineering
- 50% reduction in innovation time-to-market
Christopher, Process Engineer at Polar, used Altara to generate data-backed hypotheses in minutes, pressure-test promising ideas, and validate the strongest ones on the plant floor. He described using Altara as “having a team of 10 more engineers across all functions next to me running ideas, simulations, and analyses in parallel.”
And this was just one project within a broader deployment across Polar’s R&D, quality, manufacturing, and process engineering teams. As Polar CEO Michael Edelman puts it: "With Altara as part of the team, our pace of innovation is faster than it’s ever been."
We're excited to continue driving outcomes like these for our customers across the semiconductor, advanced materials, and battery industries.
↓ Read the full case study below.
Advay Pal retweeted
Excited to announce Agent Mode.
1/ Right now, agents are struggling to use websites on the internet. As a result, personal agents take too long to reply, and don’t get the information they need.
Advay Pal retweeted
1/ Can AI help doctors make more informed decisions based on patient activity and behavior over the course of a hospital stay?
Today we introduce TRACER: a training-free multimodal harness that enables LLMs and VLMs to track patient activity using bedside video and health records.
Blogpost: ai.stanford.edu/blog/tracer/
Paper: arxiv.org/abs/2610.21266
Advay Pal retweeted
Introducing Decision-Bench, a new industry standard for measuring model capabilities in making ROI-positive decisions on messy enterprise data.
Every enterprise has slowed AI investments due to no ROI; AIs still fail at complex decisions, and pricing second order effects.
Advay Pal retweeted
Today we are finally coming out of stealth with a few announcements:
- Founding @thirdmomentres with my friend Jeffrey Yan: automatically improving inference endpoints (with focus on data research).
- Fundraise at $250M valuation earlier this Summer with investors including @neo (@apartovi ), @Mascobot , @adamdangelo and more.
- 0 -> 8 figures in revenue within ~6 months of founding the company. In just this past month, we won $10M+ in contracts.
- Hiring ~15 more engineers in the next 45 days. Currently based in SF, opening a NYC office soon.
Jeffrey and I are building Third Moment to do two things: (1) make it effortless for enterprises to deploy state-of-the-art AI tailored to their workflows, and (2) build best-in-class RL environments that push the frontier of model capabilities. The design of a reinforcement learning environment has an underappreciated impact on the resulting behaviors of the models trained on it.
If you are interested in pushing the frontier of ML systems and data research, reach out to us ([email protected]). Check out OpenAI’s blog post on some of our experiments in the comments.
Advay Pal retweeted
Significant update from Jack Clark this morning. His RSI timelines have moved up. He now says it is likely that full RSI will become possible in 2027, and that creativity was the last piece. I think we are less than six months away from closing the loop.
Most of the recent math results were not reached by a swarm, the way Navier–Stokes was. I think that part got lost in the excitement around the release. OpenAI's unnamed internal model, which I'm going to call Aeon, reached most of them in one shot, from a single prompt, with no interruptions. This model did not even exist before the end of August, and it is still training. Notice how the returns are not dropping off? That chart is from a month ago. It is a log scale. What does Aeon look like now?
On average, each result used three hours of thinking. Aeon was given 4000 problems to work on by OpenAI. How many more have been solved in the last three days? That number is not zero.
It's like hearing notes in a song that is slowly rising. I don't think Pacing the Frontier was just about Hugging Face or hacking. They've seen how close we are to closing the loop, and as the hour draws near, their resolve begins to quaver. The last piece was model creativity, and I think that threshold was crossed internally by Anthropic and OpenAI in September. You see it in Opus 5.5, which gets it from Fable 5.5. You see it in the math results from Aeon.
All that was needed to start the event was the ability for models to think of novel ways to improve themselves. That was the last piece. This is directly analogous to the ability to think of strange, alien ways to solve math problems: solutions so inhuman that they are difficult to express in existing human terms, so the proof winds up incomprehensible. I think these same kinds of alien solutions are now being applied to model improvements internally. That's what all this recent consternation is really about. They see the invisible frontier. And they see what is about to happen.
Advay Pal retweeted
Frontier models still struggle to accurately understand human video data for robotics
The biggest gap is actions: GPT-Astra hits 71.5% accuracy, Meckas custom trained model achieves 94%
@tkkong + @philhchen evaluates Meckas video understanding model against the frontier below
Today, we're sharing results from our partnership with @MeckaAI to see if frontier models can label egocentric videos for robot learning.
We had five frontier models caption 200 clips with human grading.
GPT-6 Astra which performed the best got 43.5% of clips fully correct.
Advay Pal retweeted
I spoke with @AndrewRSorkin on CNBC "Squawk Box" about the right way to make sure AI progress is good for society.
cnbc.com/video/2026/10/08/ai…
(If you agree, use my voting guide at sway.co/ai to elect members of Congress and state AGs who align with AI accountability.)
