Khaled Saab (@_khaledsaab) header
Khaled Saab (@_khaledsaab) profile picture
California, USA
Joined March 2019
Photo by Khaled Saab (@_khaledsaab)Photo by Khaled Saab (@_khaledsaab)Photo by Khaled Saab (@_khaledsaab)Photo by Khaled Saab (@_khaledsaab)Photo by Khaled Saab (@_khaledsaab)Photo by Khaled Saab (@_khaledsaab)Photo by Khaled Saab (@_khaledsaab)Photo by Khaled Saab (@_khaledsaab)Photo by Khaled Saab (@_khaledsaab)Photo by Khaled Saab (@_khaledsaab)
Khaled Saab retweeted
I've always believed that intelligence should come from simple, general building blocks that learn their own abstractions directly from raw data; in other words, the bitter lesson. I think we're finally seeing a real step towards AI that can learn from all data. Language modeling on raw bytes is a microcosm of this ideal, and to me it's a central problem in AI: if we can't build LLMs that learn end-to-end from bytes, how will we handle the rest of the world's data that we understand even less? Last year we introduced hierarchical networks (H-Nets) as a first step. Our new version finds up to 3-5x efficiency gains in controlled scaling law experiments, and at larger scale matches an open-source model trained on 11T tokens with ~10x less data and compute. The posts below walk through what we found and why we think it works.
261
826
288
7,388
1,044,146
Khaled Saab retweeted
More than 300 million people use ChatGPT for health and wellness questions each week. OpenAI’s Karan Singhal thinks that’s only the beginning. Karan joins a16z's Jay Rughani to explain how OpenAI is building health intelligence, the advantages of using generalized models for health applications, how AI could lower the cost of healthcare, and more. 00:00 Intro 01:00 How OpenAI is building health intelligence 07:57 A doctor in your pocket 14:20 Creating trust with users 19:44 Medical consensus and its 5–17 year lag 28:03 AI for mental health 30:59 How AI enables the healthcare ecosystem 38:21 A "move 37" for health 43:39 Driving down the cost of care 50:27 Procedural care and what's next YouTube: youtu.be/9n0MI4D7TCM @thekaransinghal @JayRughani
17
32
14
196
117,255
Khaled Saab retweeted
Delighted to see our real-world clinical study of a conversational diagnostic AI system published in @TheLancet, alongside coverage in @nytimes last week. Looking back, AI in medicine might feel inevitable but it actually takes deliberate, uncertain leaps to get there. I have been fortunate to be part of two such moments. The first was early 2022 (pre ChatGPT) when @alan_karthi and I got access to our first @Google LaMDA and PaLM models. That convinced us to go all in on LLMs in medicine, sparking the Google Brain moonshot leading to Med-PaLM. The second moment leading directly to this Lancet study, was late summer 2023. @taotu831 showed me the early AMIE models he was training with our self-play loop, targeting our upcoming OSCE study with patient actors, later published in @Nature 2025. At that exact time, I was navigating a personal health condition. Testing myself, the consultation the AI performed was so natural and clinically sound that I was convinced it was ready for the real world. So I reached out to Alan and together with the team we asked how we could test this in the real world quickly, safely and responsibly. We needed a prospective clinical trial with a safety endpoint and all roads led to @AdamRodmanMD at @bethisraellahey @harvardmed . Adam was incredibly gracious with his time kicking off our first discussions in the fall of 2023. But clinical trials take a long time; years of painstaking work. There are so many critical people and moments that make them happen, far beyond the glitzy PR materials and the main authors. - For example, a discussion with @greg_corrado sparked the design of the oversight setup underpinning the safety layer in the study. - Kavita Kulkarni (now at BioHub) spent months negotiating the initial contracts with BIDMC; it was uncharted territory for both organizations. - @_khaledsaab (now rocking at OpenAI!) working tirelessly to improve the AI, together with @taotu831, @apalepu13, @RyutaroTanno, @ckbjimmy, @valentinlievin, @elahevedadi and many more folks. - @Mysiak, Alessio Orlandi, Petar Sirkovic and team sharing the seamless serving infrastructure they built for Co-Scientist, which we could re-use for much of this study. - @SSaraMahdavi making sure the study team was well organized :) This is just to name a few. Prospective trials are a massive team sport and I am proud of what our teams achieved. While the AI's clinical reasoning was strong, seeing the patient trust, safety, and satisfaction data, with ZERO safety stops from our supervising physicians, is what truly stands out. At the same time, this is still a stepping stone. We are moving from AI as a tool to an active care team member - a guardian angel amplifying clinicians and staying with patients throughout their health journey. More to come soon, and as always, a pleasure to work with so many great folks across @GoogleDeepMind , @GoogleResearch and @bethisraellahey . Paper - thelancet.com/journals/lance… NYT - nytimes.com/2026/10/08/techn…
10
16
70
4,036
GPT-6 Luna and Sol are a step towards health intelligence too cheap to meter. ♥️ These models push the frontier of performance per dollar on HealthBench Professional. GPT-6 Luna scores higher than GPT-5.6 Sol–our best model a few weeks ago–at their highest reasoning efforts, despite GPT-6 Luna being ~34x cheaper: about a third of a cent per response. HealthBench Professional consists of real clinician tasks, including care consultations, medical documentation, and research. This evaluation is designed to be challenging and focuses on especially complex and rare situations. I’m excited for what abundant health intelligence can unlock for patients and clinicians.
7
11
2
102
11,658
Verified U.S. clinicians can get free GPT-6 Astra (Pro) access today via ChatGPT for Clinicians. Sign up: chatgpt.com/plans/clinicians…
204
258
140
2,744
1,260,984
Khaled Saab retweeted
following up after astra launch week and the navier-stokes announcement i wanted to share a little bit about how we think about and build for the over 1 billion people who use chatgpt every week, most of whom use it for free. we’re on a mission to distribute the benefits of agi, and i think this is the most impactful way to do that: giving people as much access as we can safely and feasibly distribute. we want to scale the utility that everyone gets from ai. over the past 6 months, we've been working to bring the greatest possible capabilities to the broadest possible set of users, including both free and paying users. of course there are some practical limitations and we can’t ship the largest models to everyone, but since march (gpt-5.3 instant), we’ve improved the default chatgpt experience meaningfully: - responses with a major factual error are down 65%, and they are down 72% in high stakes categories like finance - gpt-5.6 sol at instant and gpt-5.6 luna at medium are smarter than o3 at high reasoning effort, our frontier reasoning model from 17 months ago, while being considerably faster, 30%+ faster time to last token (evaluated via GPQA diamond) - chatgpt is far more usable and enjoyable while bringing extreme sycophancy down 80% and overall sycophancy down 83% - on our high-stakes medical eval, chatgpt produced 83% fewer answers flagged for hallucinations, resulting in hundreds of millions of better health conversations per month - we're seeing people find more ways to use chatgpt. for a recently joined cohort, we saw the number of use cases pursued at least three times weekly go up 20-30% not only have we improved the default chatgpt experience dramatically, we’ve also expanded access significantly for all. free users now: - have access to unlimited text chats - can use higher reasoning effort - now have access to automations, letting them schedule tasks to happen later - have access to better memory through dreaming. this allows chat to personalize much more effectively we’re always looking for ways to expand access to the billion+ people who use our product weekly. this is the master plan: bring the default experience as close to the capability frontier as possible, expand access to everyone, making intelligence maximally abundant, and make our product much more usable. this last point, usability, is something we’ve been focusing on as we bring personal agi to everyone. the current top of the line capabilities are staggering, but only the savviest users and insiders know how to use them. our vision is of a progressive, model assisted, onboarding to agi. chatgpt will guide people from all walks of life to the highest utility applications of ai for them. this really is the only way – we think technology that is hard to learn ultimately fails to reach a broad audience. your personal agi will just understand your goals, and automatically use the right set of features and capabilities to get things done for you lots more to come soon. onwards to scaling utility for all
52
77
10
811
124,380
Khaled Saab retweeted
I wrote about the state of AI, why I’m concerned about the next few years, and the choices we need to make to keep the future in humanity’s hands. An Alien Mind: openai.com/index/an-alien-mi…
958
2,508
1,092
15,249
7,699,251
Khaled Saab retweeted
GPT-6 Astra, trained on ~100K+ NVIDIA Grace Blackwell NVLink72. From ChatGPT to o1 to Astra in 4 years. AGI has arrived. Congratulations @OpenAI team. 400K GPUs coming online next.
1,844
4,717
2,827
41,643
12,301,211
Khaled Saab retweeted
GPT-6 Astra represents a step-function change in model capability for interactive reasoning problems. It scores 66% on ARC-AGI-3 using our standard harness, and nearly 100% with a continuous conversation harness and custom compaction, at a cost of roughly $360 per game. In fact, the continuous harness version significantly outperforms our human baseline in action efficiency across almost all levels. When we examined the reasoning chains to understand how the model operates, we found it performing highly efficient, on-the-fly symbolic world modeling for each game and level. It goes as far as developing its own shorthand DSL to represent in-game situations -- essentially a game-specific algebraic notation. Overall, Astra exhibits symbolic modeling behaviors we had previously only seen with sophisticated harnesses -- so harness capabilities are increasingly shifting into the model itself. We see Astra as a major breakthrough in model intelligence. Read our post on Astra and what these results mean: arcprize.org/blog/astra
193
867
258
6,921
987,174
Khaled Saab retweeted
This is GPT-6 Astra. Anything you can do on a computer, Astra can do for you. Fast.
9,104
35,091
26,628
337,906
139,469,807
Khaled Saab retweeted
I want to prevent a race into unmonitorability kicked off by confused reporting. The depth of the computation graph for our present frontier models, including Astra, is within a factor of two of GPT-4. OpenAI has worked to preserve and utilize chain-of-thought monitoring since our very first reasoning models. We deeply care about this technique, as it can give us a view into how model alignment generalizes from its training distribution. I do think it is fragile and unfortunately trending in a negative direction, for reasons not contingent on architecture changes that I will write about soon. But there are things we can do to strengthen it, and it's a core goal of our current research program.
270
510
195
6,427
1,658,076
Today, we’re bringing ChatGPT closer to the systems, information, and workflows healthcare teams already rely on. ♥️ We’re introducing a new EHR integration to connect supported Epic environments to ChatGPT and a plugin connecting to nine additional industry data sources.
122
206
128
2,138
997,350
Khaled Saab retweeted
one of the most beautiful things about OpenAI is that every employee really has a voice. i wanted to capture what it feels like to work here, what our mission means to me, and why you should join us. so i made this video with Codex, shared it with the team, and they felt it was worth producing and sharing with the world. this is our mission. and it’s why i’m here.
297
244
98
3,078
2,383,696
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.
92
55
15
705
168,566
Khaled Saab retweeted
Health in ChatGPT is starting to roll out to U.S. users. You can securely connect Apple Health and supported medical records to understand your information in context, track what has changed, and have more informed conversations. openai.com/index/health-in-c…
432
390
270
4,544
1,282,567
♥️ GPT-5.6 is a major step forward for health, both at the frontier and at cost. These models push the frontier of performance per dollar, bringing the best health intelligence to all. The smallest variant, GPT-5.6 Luna, evaluated at the lowest reasoning effort, outperforms GPT-5.5 at the highest reasoning effort–despite costing 25x less. The largest variant, GPT-5.6 Sol, sets a new high bar at cost. Another especially cool result: physicians found fewer flaws in GPT-5.6 responses than physician-written responses. We collected diverse tasks that remain difficult for recent OpenAI models, across patient-facing and clinician-facing use cases. We asked speciality-matched physicians to write responses to these tasks with unlimited time and web access. We then asked other physicians to compare responses side-by-side, blinded to their source. Physicians were asked to comment on areas of improvement across five axes: accuracy, communication, completeness, instruction following, and health decision helpfulness. We then reported the fraction of responses across sources rated perfectly across all axes, across 20,000 total axis ratings. GPT-5.6 Sol appeared strongest, although all GPT-5.6 models performed significantly better than physicians.
GPT-5.6 is a major step forward for health intelligence. Across the lineup, we’re delivering stronger performance at lower cost: GPT-5.6 Luna outperforms GPT-5.5 at its highest reasoning setting while costing 25x less. Together, these advances raise quality while making advanced models accessible to more people globally.
57
128
35
951
1,399,564
One of the ways scaling test-time compute can benefit people most: have reasoning models think really hard about rare undiagnosed diseases. Today we’re sharing published evidence that this can work, in some of the most difficult pediatric cases!
Together with researchers at Boston Children’s Hospital and Harvard, we published a study in NEJM AI showing how o3 Deep Research helped clinicians revisit previously unsolved rare pediatric disease cases, and find answers for families who had waited years.
108
90
10
827
445,737
Our paper “Towards Conversational AI for Disease Management” is out in @Nature! We ran a blinded study comparing our multi-agent AMIE system to 21 primary care physicians to test longitudinal patient care management. 🔗 nature.com/articles/s41586-0… @GoogleDeepMind @GoogleResearch
2
14
1
66
4,484
Khaled Saab retweeted
GPT-5.5 early access results have been impressive. 25% lift in clinical quality. 30% less verbose. We orchestrate hundreds of AI tasks, some powered by our proprietary data flywheels, others by frontier models like GPT-5.5. We test rigorously for clinical accuracy, completeness, reasoning, real-world because benchmarking is a big deal in healthcare. Grateful to the OpenAI team for the early access and partnership.
4
11
61
4,640