Jeff Dean(@JeffDean)

A nice aspect of our new Gemini 3.6 Flash model is that it is much more token efficient than our 3.5...

Jeff Dean (@JeffDean) — Tue, 21 Jul 2026 15:37:20 GMT

A nice aspect of our new Gemini 3.6 Flash model is that it is much more token efficient than our 3.5 Flash model.   Here's a side-by-side demonstration of that.

Nice work by everyone who worked on this model and release!

Your browser does not support the video tag.

Google DeepMind@GoogleDeepMind

We’re rolling out three new models to make AI agents faster, smarter, and cheaper at scale:

🔵 Gemini 3.6 Flash: It uses fewer tokens than 3.5 Flash to deliver higher quality work at the exact same cost.

🔵 Gemini 3.5 Flash-Lite: A fast, cost-effective option for everyday tasks like processing documents and agentic search.

🔵 Gemini 3.5 Flash Cyber: A cybersecurity model built to find and patch critical software vulnerabilities.

💬84🔄80❤️958👀149011📊145

Yesterday I was fortunate enough to go to my first-ever World Cup game, with my long-time colleagues...

Jeff Dean (@JeffDean) — Mon, 20 Jul 2026 14:21:30 GMT

Yesterday I was fortunate enough to go to my first-ever World Cup game, with my long-time colleagues/friends @OriolVinyalsML and @quocleix and their spouses).  We were sitting in the corner area and had a quite good view of the goal!

I present to you a 43 second multi-part drama filled with emotion:

The initial promising-looking cross coming in, but looking overhit

Nico Williams cleverly knocking it down at the back post into a dangerous area

Ferran Torres striking it cleanly into the roof of the net

The crowd rising as one (forcing me to stand up as well)

The elation of the Spanish players racing off the bench to celebrate

The dejection of the Argentinian players, their defense having finally been breached in extra time

The elation of my Spanish colleague Oriol and his wife Meire next to me (he and I are both Barça fans, so it was nice to see a Barça player score the winning goal)

The entire stadium reacting

Whew!

Your browser does not support the video tag.

💬58🔄46❤️2043👀164879📊261

The video from the commencement speech I gave a couple of weeks ago at @uwcse is now up. Congratula...

Jeff Dean (@JeffDean) — Tue, 30 Jun 2026 23:22:11 GMT

The video from the commencement speech I gave a couple of weeks ago at @uwcse is now up.  Congratulations again to all the graduates, and thanks for having me, @MBalazinska!

youtu.be/Pvkmwh7Mito?si… (video is of whole ceremony: my part starts around the 3 minute mark)

Jeff Dean@JeffDean

I enjoyed giving the commencement address at the University of Washington Allen School @uwcse graduation this evening.  So many happy students and their families and friends!

Congratulations to all the graduates of the class of 2026! 🎓

Thanks for inviting me, @MBalazinska!

💬20🔄9❤️185👀56842📊31

My @Google colleagues @NormJouppi, Sridhar Lakshmanamurthy, Cliff Young, and David Patterson recentl...

Jeff Dean (@JeffDean) — Thu, 18 Jun 2026 18:51:38 GMT

My @Google colleagues @NormJouppi, Sridhar Lakshmanamurthy, Cliff Young, and David Patterson recently wrote a paper that will appear in the July/August 2026 edition of @ieeemicro titled "Google's Training Supercomputers from TPU v2 to Ironwood: Architectural Stability, Scale, Resilience, Power Efficiency, and Sustainability Across Five Generations".  It's chock full of interesting data about the evolution of TPU chip generations, as well as how workloads at Google have transformed over time (hint: lots more transformer-based models!), and how the generations have gotten ~30X more energy efficient per flop.

Lots of changes over these generations:
Air cooling in TPUv2 to water cooling in TPUv3 onwards
2D to 3D torus-based interconnects
30X improvement TFLOPS/Watt 
256 chips (TPUv2) to 9216 chips (Ironwood) per pod

Read the full paper: arxiv.org/abs/2606.15870

💬30🔄159❤️1150👀118300📊250

I really enjoyed this game today. Vozinha and the whole Cape Verde team were amazing against Spain!

Jeff Dean (@JeffDean) — Tue, 16 Jun 2026 03:36:00 GMT

I really enjoyed this game today.  Vozinha and the whole Cape Verde team were amazing against Spain!

💬17🔄3❤️165👀21563📊29

A good essay by @pgasawa and @profjoeyg on a more nuanced view of AI advances.

Jeff Dean (@JeffDean) — Mon, 15 Jun 2026 18:36:25 GMT

A good essay by @pgasawa and @profjoeyg on a more nuanced view of AI advances.

Parth Asawa@pgasawa

The AI community seems to increasingly be heading towards a polarized world when discussing safety and consolidated power. I see this discourse as a false dichotomy, so @profjoeyg and I wrote an essay on how we need to change the conversation (link below).

💬15🔄19❤️92👀32849📊27

@uwcse @MBalazinska Here's a nice @GeekWire article by @ToddBishop about the graduation, including a...

Jeff Dean (@JeffDean) — Sat, 13 Jun 2026 21:47:13 GMT

@uwcse@MBalazinska Here's a nice @GeekWire article by @ToddBishop about the graduation, including a few more snippets of what I said in my talk.

geekwire.com/2026/what-a-lo…

💬0🔄4❤️30👀11089📊5

@uwcse @MBalazinska Here are a few of the pieces of advice that I shared:

Jeff Dean (@JeffDean) — Sat, 13 Jun 2026 05:14:48 GMT

@uwcse@MBalazinska Here are a few of the pieces of advice that I shared:

💬23🔄148❤️1073👀76886📊245

I enjoyed giving the commencement address at the University of Washington Allen School @uwcse gradua...

Jeff Dean (@JeffDean) — Sat, 13 Jun 2026 05:04:10 GMT

I enjoyed giving the commencement address at the University of Washington Allen School @uwcse graduation this evening.  So many happy students and their families and friends!

Congratulations to all the graduates of the class of 2026! 🎓

Thanks for inviting me, @MBalazinska!

💬20🔄26❤️735👀117557📊89

People replace their phones every ~4 yrs. This means there are hundreds of millions of old phones d...

Jeff Dean (@JeffDean) — Sat, 13 Jun 2026 04:17:25 GMT

People replace their phones every ~4 yrs.  This means there are hundreds of millions of old phones discarded each year that are still perfectly usable as computing devices.  @Google in collabration with @UCSD is exploring how to turn these old phones into cloud-computing “phone clusters”.  Putting phones back in service in this way can directly reduce the environmental footprint of computing by avoiding the need for further raw material extraction, and taking advantage of the embodied carbon already incurred from manufacturing these devices, and modern phones actually are already quite powerful computers.  Read more in the blog below ⬇️

Google Research@GoogleResearch

Today on the blog, we discuss a pathway for the second life of phones through the exploration of “phone cluster computing”, which can directly reduce the environmental footprint of computing by avoiding the need for further raw material extraction. More →goo.gle/4aJe5vO

💬160🔄529❤️5292👀652700📊780

Quite interesting thread on capabilities of real biological neurons (spoiler: they're way more capab...

Jeff Dean (@JeffDean) — Fri, 12 Jun 2026 16:30:06 GMT

Quite interesting thread on capabilities of real biological neurons (spoiler: they're way more capable than classical artificial neurons in a perceptron) .  Nice work @IdoAizenbud and collaborators!

Ido Aizenbud@IdoAizenbud

What can a neuron compute?

Real biological neurons are complex, but how capable are they?

Using a new method, we found that a single cortical neuron can classify cats vs dogs, recognize spoken words, and solve 10-bit parity, all tasks thought to require entire networks. (1/15)

💬28🔄83❤️670👀100404📊130

Denying entry to a Somali soccer official selected as one of the World Cup referees is quite shamefu...

Jeff Dean (@JeffDean) — Tue, 09 Jun 2026 23:59:34 GMT

Denying entry to a Somali soccer official selected as one of the World Cup referees is quite shameful. The whole point of the World Cup is as a spirited athletic competition that brings us together, and allowing him to officiate is obviously the right thing to do.

BBC Sport@BBCSport

Somali referee Omar Artan, who was set to be the first from his country to officiate at the World Cup finals, has been denied entry to the United States.

💬92🔄83❤️1043👀88645📊148

You can read more about this here: https://t.co/mDg1zOOLuD

Jeff Dean (@JeffDean) — Tue, 09 Jun 2026 17:34:42 GMT

You can read more about this here: blog.google/innovation-and…

💬3🔄10❤️28👀7895📊8

Speech translation has been one of the longest-running ML efforts at Google, and we’ve come a long w...

Jeff Dean (@JeffDean) — Tue, 09 Jun 2026 17:34:13 GMT

Speech translation has been one of the longest-running ML efforts at Google, and we’ve come a long way.  Gemini 3.5 Live Translate is our latest speech-to-speech model, supporting 70+ languages.  It enables more natural conversations across languages in everyday products and apps.

Here’s an example of how partners at @InsideGrab are helping connect travelers with drivers. 🚗

Rolling out in Google Translate and via the Live API in @GoogleAIStudio.

Your browser does not support the video tag.

💬29🔄44❤️372👀35428📊77

Congratulations, Ayanna Howard @robotsmarts! Spelman has made an excellent choice to have you as it...

Jeff Dean (@JeffDean) — Sat, 06 Jun 2026 04:07:03 GMT

Congratulations, Ayanna Howard @robotsmarts!  Spelman has made an excellent choice to have you as its next president!

Spelman College@SpelmanCollege

Today, Spelman College proudly announces Dr. Ayanna Howard as its 12th President. This historic transition marks the next chapter in Spelman's legacy of excellence, empowering future generations of Black women to lead, innovate, and shape a rapidly changing world. A pioneering roboticist and higher education leader, Dr. Howard brings a bold vision for Spelman's future while honoring the mission and values that have defined the College for more than 140 years. Read more: spelman.edu/news/2026/06/s…

💬11🔄12❤️84👀27305📊15

Check out our Gemma 4 12B model: it's a super capable open weights model that can run directly on yo...

Jeff Dean (@JeffDean) — Thu, 04 Jun 2026 02:00:37 GMT

Check out our Gemma 4 12B model: it's a super capable open weights model that can run directly on your laptop.

Google Gemma@googlegemma

Meet Gemma 4 12B!

A unified, encoder-free multimodal model designed to bring high-performance intelligence directly to your laptop, and released under an Apache 2.0 license.

Bridging the gap between edge efficiency and advanced reasoning. Here is what’s new with Gemma 4 12B: 👇

💬32🔄47❤️578👀56240📊76

This is great to see. The scientific community & the general public should be in the habit of c...

Jeff Dean (@JeffDean) — Wed, 03 Jun 2026 03:46:41 GMT

This is great to see.  The scientific community & the general public should be in the habit of celebrating amazing scientific achievements with standing ovations 🎉

💬24🔄49❤️674👀84428📊96

Thanks for a great @twominutepapers conversation, Károly!

Jeff Dean (@JeffDean) — Tue, 02 Jun 2026 01:08:56 GMT

Thanks for a great @twominutepapers conversation, Károly!

Two Minute Papers@twominutepapers

It was really fun to chat with the legendary @JeffDean. I asked him about things that only he may know the answer to. 🙂 Link to full video in thread.

💬10🔄16❤️123👀29998📊26

I enjoyed this chat immensely with @OfficialLoganK and my fellow Gemini leads @OriolVinyalsML and @N...

Jeff Dean (@JeffDean) — Fri, 29 May 2026 21:39:16 GMT

I enjoyed this chat immensely with @OfficialLoganK and my fellow Gemini leads @OriolVinyalsML and @NoamShazeer and @koraykv.

Logan Kilpatrick@OfficialLoganK

My conversation with @JeffDean, @koraykv, @NoamShazeer, and @OriolVinyalsML (the Gemini co-leads) about the current state of Gemini, how we got here, and where we go next : )

Your browser does not support the video tag.

💬25🔄33❤️395👀57156📊75

Four questions in under 2 minutes. Thanks to Google's social team for suggesting this little interlu...

Jeff Dean (@JeffDean) — Wed, 20 May 2026 20:17:12 GMT

Four questions in under 2 minutes. Thanks to Google's social team for suggesting this little interlude at #googleio yesterday!

Google@Google

4 questions with @JeffDean (Chief Scientist, Google and one of the authors of the famous "cat paper" about unsupervised learning and neural networks in 2012) in under 2 minutes at #GoogleIO:

Your browser does not support the video tag.

💬18🔄15❤️218👀36311📊32

Highly capable models that are fast are super important. Our new Gemini 3.5 Flash model is a great ...

Jeff Dean (@JeffDean) — Tue, 19 May 2026 21:43:59 GMT

Highly capable models that are fast are super important.  Our new Gemini 3.5 Flash model is a great mix of fast and capable.

Sundar Pichai@sundarpichai

Just off stage at #GoogleIO, some highlights from this morning 🧵

Gemini 3.5 Flash is available today for everyone in @antigravity and across our products and APIs.

Compared to 3.1 Pro, 3.5 Flash is better across almost all benchmarks with huge progress in coding. It’s also comparable to the best models but very fast (4x faster tokens/ second than other frontier models). And when looking at the intelligence versus output speed, it’s in a league of its own in the top right quadrant.

💬44🔄20❤️266👀30010📊52

Beautiful day at #GoogleIO packed full of Gemini announcements, with my Gemini team colleagues @Orio...

Jeff Dean (@JeffDean) — Tue, 19 May 2026 18:27:50 GMT

Beautiful day at #GoogleIO packed full of Gemini announcements, with my Gemini team colleagues @OriolVinyalsML and @borgeaud_s!

Take a sip of water every time you hear "Gemini"!

Oriol Vinyals@OriolVinyalsML

This is my third Google I/O in a row. Some things have changed, some (like the amazing colleagues that were there since day 0) haven't. @JeffDean@borgeaud_s

💬18🔄12❤️351👀26331📊38

3/ Gemini 3.5 Flash is rolling out globally today. On behalf of the entire Gemini team, we're excit...

Jeff Dean (@JeffDean) — Tue, 19 May 2026 17:49:55 GMT

3/ Gemini 3.5 Flash is rolling out globally today.  On behalf of the entire Gemini team, we're excited by what you'll be able to do with this model!

Read more here: blog.google/innovation-and…

💬2🔄8❤️121👀14191📊15

2/ Check out how Gemini 3.5 Flash instantly digests dense academic papers and autonomously codes a f...

Jeff Dean (@JeffDean) — Tue, 19 May 2026 17:46:45 GMT

2/
Check out how Gemini 3.5 Flash instantly digests dense academic papers and autonomously codes a fully interactive, visual website explaining the intricacies of the research.  It's an incredible stress test that seamlessly merges massive long context, deep reasoning, complex coding, and ultra-low latency.

It really helps you distill papers down to their essence and aid your understanding!

Your browser does not support the video tag.

💬6🔄26❤️273👀88439📊63

1/ Today at #GoogleIO, we’re releasing Gemini 3.5, our latest family of models combining frontier in...

Jeff Dean (@JeffDean) — Tue, 19 May 2026 17:45:39 GMT

1/ Today at #GoogleIO, we’re releasing Gemini 3.5, our latest family of models combining frontier intelligence with action.

We’re starting by releasing 3.5 Flash, which is built to help you execute complex, long-horizon agentic workflows.

Gemini 3.5 Flash is our strongest model for coding and agent yet.It outscores 3.1 Pro on agentic and coding benchmarks like Terminal-Bench and MCP Atlas, while running 4x faster than other frontier models.

Used in Google Antigravity, 3.5 Flash is even further optimized to be up to 12x faster. It’s a powerful engine to deploy sub-agents that collaborate, run high-frequency iterative loops, and solve real-world problems at scale.

Some highlights we’re excited about 🔽

💬85🔄195❤️1486👀135855📊283

Those early whole company @Google ski trips were fun! I'm eighth from the left in the top row, wear...

Jeff Dean (@JeffDean) — Fri, 15 May 2026 04:57:53 GMT

Those early whole company @Google ski trips were fun!  I'm eighth from the left in the top row, wearing the white bathrobe (ninth if you count Susan's baby, Ari)

Paul Buchheit@paultoo

Don't just start a company
Start a cult

💬52🔄39❤️2014👀231160📊203

Great to see @percyliang as a keynote speaker at #cais2026!

Jeff Dean (@JeffDean) — Tue, 12 May 2026 20:15:45 GMT

Great to see @percyliang as a keynote speaker at #cais2026!

ACM Conference on AI and Agentic Systems@CAISconf

🎤 Keynote announcement: 
@percyliang (Percy Liang), Professor of Computer Science at @Stanford, founding director of the Center for Research on Foundation Models, and co-founder of @togethercompute, is keynoting#CAIS20266.

Percy's HELM framework set the standard for holistic evaluation of language models, and his Foundation Model Transparency Index (now in its third year) put every major AI lab on notice for what they do and don't disclose. His current work on Marin takes this further: an open lab where every experiment, successful or not, is public from day one.

This one's going to be good.

San Jose · May 26–29caisconf.orgm

💬11🔄19❤️224👀48282📊33

Bleh. Talks. But dishes also works. The keynotes are like a multicourse tasting menu!

Jeff Dean (@JeffDean) — Wed, 06 May 2026 04:26:15 GMT

Bleh.  Talks.  But dishes also works.  The keynotes are like a multicourse tasting menu!

💬2🔄2❤️24👀10993📊5

Great to see my friend @andykonwinski as one of the keynote dishes at the first @CAISconf later this...

Jeff Dean (@JeffDean) — Wed, 06 May 2026 01:59:32 GMT

Great to see my friend @andykonwinski as one of the keynote dishes at the first @CAISconf later this month!

ACM Conference on AI and Agentic Systems@CAISconf

📢Keynote announcement:

@andykonwinski (Andy Konwinski), co-founder of Databricks and Perplexity AI, founder of @LaudeInstitute, is keynoting#CAIS20266.

Laude's "Ship Your Research" mission funds open-source AI research through its Moonshots, Slingshots, and Open Frontier programs. Case in point: Terminal-Bench, a Laude-backed agent benchmark, reached Anthropic's Claude 4 model card 126 days after inception and has since become the industry-wide standard for measuring command line performance.

We're thrilled to have him.

San Jose · May 26–29caisconf.orgm

💬12🔄25❤️259👀53563📊38

Google Translate is turning 20! 🎉. There are 20 fun facts and tips in the thread below. Translate ...

Jeff Dean (@JeffDean) — Tue, 28 Apr 2026 20:16:21 GMT

Google Translate is turning 20! 🎉. There are 20 fun facts and tips in the thread below.

Translate is one of my favorite Google products because it brings us all closer together!

I've been involved with a couple of things over the years. The first was our deployment of the initial system in 2006, which provided a huge leap forward in quality because it used a much larger 5-gram language model trained on trillions of words of text (indeed, probably the first trillion token language model training in the world: paper has some nice heads showing scaling-law-like quality improvement from scaling to more data/compute).

See "Large Language Models in Machine Translation", Thorsten Brants, Ashok C. Popat, Peng Xu, Franz J. Och and Jeffrey Dean,aclanthology.org/D07-1090/j

The second major collaboration was in 2016 when we moved Translate over from a statistical machine translation approach to using deep neural networks.  This approach relied on two key innovations.  The first was Google's work on Sequence-to-Sequence models arxiv.org/abs/1409.3215V).  The second was our development of TPUs, custom cups that improved the performance of inference for deep neural networks by 30-80X over existing CPUs and GPUs of the day (and reduced latency by 15-30X).  This made launching compute-intensive language model services like Translate feasible for hundreds of millions of users. See "In-Datacenter Performance Analysis of a Tensor Processing Unit",  Norman P. Jouppi et al. arxiv.org/abs/1704.04760O

GNMT paper:
"Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation",  Yonghui Wu, Mike Schuster, Zhifeng Chen, Quoc V. Le, Mohammad Norouzi, Wolfgang Macherey, Maxim Krikun, Yuan Cao, Qin Gao, Klaus Macherey, Jeff Klingner, Apurva Shah, Melvin Johnson, Xiaobing Liu, Łukasz Kaiser, Stephan Gouws, Yoshikiyo Kato, Taku Kudo, Hideto Kazawa, Keith Stevens, George Kurian, Nishant Patil, Wei Wang, Cliff Young, Jason Smith, Jason Riesa, Alex Rudnick, Oriol Vinyals, Greg Corrado, Macduff Hughes, and Jeffrey Dean,arxiv.org/abs/1609.08144M

Most recently, we have advanced Translate further using Gemini models.

Each of these advances relied on research that have major quality leaps over the existing status quo translation approaches, bringing better quality and connectedness to all of our Translate users! 🎉

💬32🔄68❤️462👀184584📊99

The video of my conversation with Amin Vahdat, @gilbert, and @djrosent at Cloud Next last week is no...

Jeff Dean (@JeffDean) — Mon, 27 Apr 2026 13:40:08 GMT

The video of my conversation with Amin Vahdat, @gilbert, and @djrosent at Cloud Next last week is now up. 

youtu.be/BpnJYJmbXcM?si…

Thanks for a great conversation!

Jeff Dean@JeffDean

I had a good time discussing yesterday's Google TPU v8t and v8i announcement at Cloud Next with Amin Vahdat along with @AcquiredFM hosts @gilbert
and @djrosent.

The blog post announcement has lots of details about these new chips:

blog.google/innovation-and…

Here's a thread of some particular things I'm excited about:

💬11🔄27❤️167👀40669📊33

The Arxiv for the new Decoupled DiLoCo paper is now up: https://t.co/z6MwS6Z4Ms

Jeff Dean (@JeffDean) — Fri, 24 Apr 2026 13:12:52 GMT

The Arxiv for the new Decoupled DiLoCo paper is now up: arxiv.org/abs/2604.21428…

💬3🔄3❤️75👀11238📊16

It's worth pointing out that we have been pushing on large-scale training and asynchronous technique...

Jeff Dean (@JeffDean) — Thu, 23 Apr 2026 20:15:06 GMT

It's worth pointing out that we have been pushing on large-scale training and asynchronous techniques for the last ~14 years.  Here's our NeurIPS 2012 paper where we demonstrated that this approach could be used to train very large neural networks (for the time: 30X larger than any previous neural network), and to spread the training out across thousands of machines in a fault-tolerant manner.

PDF: static.googleusercontent.com/media/research…static.googleusercontent.com/media/research…

(This paper doesn't get as much attention because we neglected to put it on Arxiv at the time: oops!)

💬7🔄4❤️84👀16893📊16

TPU 8i is co-designed with our Gemini research team to support low latency inference. Among the att...

Jeff Dean (@JeffDean) — Thu, 23 Apr 2026 20:09:30 GMT

TPU 8i is co-designed with our Gemini research team to support low latency inference.  Among the attributes that support this are large amounts of on-chip SRAM, enabling more computations to be done on chip without having to go to HBM for weights or KVCache state as often.

The boardfly network topology (see image below) offers a much lower diameter network to connect all 1152 chips in an 8i pod, by fully connecting all 4 chips on the board together, fully connecting groups of 8 boards together, and then fully connecting 36 groups of 8 boards together.  In addition, there is specialized Collectives Acceleration Engine (CAE) circuitry on each chip to offload various kinds of reductions and other global operations from the main computational portion of each chip, reducing on-chip latency by up to 5x.

Together, these features will offer very high throughput for large-scale models (including MoEs, which often require mapping onto many chips for inference), and will do so at very low latency.  This will make agentic workloads and interactive usage really shine on the TPU 8i.

💬4🔄8❤️97👀11038📊16

First, let's talk about TPU 8t, which is designed for large-scale training and inference throughput....

Jeff Dean (@JeffDean) — Thu, 23 Apr 2026 20:00:58 GMT

First, let's talk about TPU 8t, which is designed for large-scale training and inference throughput.

The pod size is increased slightly to 9600 chips, and provides ~3X the FP4 performance per pod vs. Ironwood (8t has 121 exaflops/pod vs. 42.5 exaflops/pod for Ironwood).  In addition, the ICI network bandwidth is 2X higher per chip and the scale out datacenter networking is 4X higher per chip.

Importantly, this system also offers 2X the performance / watt, continuing the trend of significant energy effiency improvements that we've had for many generations of TPUs (8t offers ~60X the performance/watt as TPU v2).

💬5🔄7❤️132👀17268📊19

I had a good time discussing yesterday's Google TPU v8t and v8i announcement at Cloud Next with Amin...

Jeff Dean (@JeffDean) — Thu, 23 Apr 2026 19:55:08 GMT

I had a good time discussing yesterday's Google TPU v8t and v8i announcement at Cloud Next with Amin Vahdat along with @AcquiredFM hosts @gilbert
and @djrosent.

The blog post announcement has lots of details about these new chips:

blog.google/innovation-and…

Here's a thread of some particular things I'm excited about:

💬18🔄81❤️518👀161797📊114

It's been a delight to provide small amounts of advice and suggestions to people working on the Deco...

Jeff Dean (@JeffDean) — Thu, 23 Apr 2026 15:41:07 GMT

It's been a delight to provide small amounts of advice and suggestions to people working on the Decoupled DiLoCo training system.  This approach enables graceful handling of failures in large scale training jobs, by allowing (N-1) / N units to proceed when one fails.

Thread ⬇️

Arthur Douillard@Ar_Douillard

The DiLoCo team at Google DeepMind and Google Research is proud to release Decoupled DiLoCo, the next frontier for resilient AI pre-training.

Decoupled DiLoCo enables training with datacenters across the world, using heterogeneous hardware, and never halting the system despite hardware failures.

💬23🔄75❤️691👀108087📊105

It was a delight to serve on the selection committee with an awesome group of committee members. We...

Jeff Dean (@JeffDean) — Wed, 15 Apr 2026 20:13:12 GMT

It was a delight to serve on the selection committee with an awesome group of committee members.  We had lots of great submissions to evaluate and excellent discussions during our selection process.

You can see the selected proposals below and learn more at laude.org/moonshots

Laude Institute@LaudeInstitute

Moonshots // ONE wouldn't exist without the people who agreed to judge it.

27 world-class researchers and experts gave serious time and thought to 125 proposals. They shaped what rigorous evaluation looks like, and held the bar high enough that some of the most decorated scientists alive didn't make the cut.

Thank you to our selection committee, chaired by Dave Patterson:
@atalwalkar, @andykonwinski, @bakkermichiel, @bradenjhancock, @chris_bail, @ChrisRytting, @davidautor, @DG_Rand, @Diyi_Yang, @emollick, @erichorvitz, @erikbryn, @ghadfield, @HannaHajishirzi, @JeffDean, @james_y_zou, John Hennessy, @JohnJumperSci, @tkalil2050, Tino Cuéllar, @tommmitchell, @Thom_Wolf, Adam Yala, Sylvia Plevritis, @YejinChoinka, @zakkohanelaude.org/moonshots

💬6🔄12❤️147👀25911📊21

In case it's not clear, you can click on the picture in the first image to learn more about Ricardo ...

Jeff Dean (@JeffDean) — Wed, 15 Apr 2026 16:13:28 GMT

In case it's not clear, you can click on the picture in the first image to learn more about Ricardo and the award: Twitter swallowed my actual link text to awards.acm.org/about/2025-bar… to turn it into the picture, I guess.

💬1🔄2❤️18👀8960📊3

Delighted to see that Ricardo Baeza-Yates (@PolarBearby) is this year's winner of the @TheOfficialAC...

Jeff Dean (@JeffDean) — Wed, 15 Apr 2026 16:12:19 GMT

Delighted to see that Ricardo Baeza-Yates (@PolarBearby) is this year's winner of the @TheOfficialACM Luiz Barroso Award (an award named in honor of my longtime Google colleague).

Ricardo is widely regarded as one of the world’s foremost researchers in information retrieval, and many of you may have read is excellent textbook Modern Information Retrieval (co-authored with Berthier Ribeiro-Neto).  He has also played a pivotal role in strengthening the Latin American computing community.

Ricardo's accomplishments and selection would make Luiz proud!

awards.acm.org/about/2025-bar…

💬13🔄24❤️231👀26375📊43

This is so awesome! When I was a kid, when the Apollo mission were still relatively fresh in the rec...

Jeff Dean (@JeffDean) — Sat, 11 Apr 2026 04:16:18 GMT

This is so awesome! When I was a kid, when the Apollo mission were still relatively fresh in the recent past, I just assumed we would send people & missions to the moon quite regularly. Hard to believe we haven't since 1972! Congrats to everyone at @NASA! 🎉

NASA@NASA

Welcome home Reid, Victor, Christina, and Jeremy! 🫶

The Artemis II astronauts have splashed down at 8:07pm ET (0007 UTC April 11), bringing their historic 10-day mission around the Moon to an end.

Your browser does not support the video tag.

💬13🔄23❤️748👀40980📊63

Congratulations to @matei_zaharia on winning this year's ACM Prize in Computing! This is very well-...

Jeff Dean (@JeffDean) — Thu, 09 Apr 2026 03:42:56 GMT

Congratulations to @matei_zaharia on winning this year's ACM Prize in Computing!  This is very well-deserved! 🎉

Association for Computing Machinery@TheOfficialACM

Congratulations to Matei Zaharia on receiving the 2025 #ACMPrize in #Computing! He is recognized for his visionary development of distributed data systems and computing infrastructure. Learn more: buff.ly/TrYHoci

💬13🔄28❤️476👀43684📊50

Great to see the reception for the very capable Gemma 4 models!

Jeff Dean (@JeffDean) — Thu, 09 Apr 2026 00:05:12 GMT

Great to see the reception for the very capable Gemma 4 models!

Sundar Pichai@sundarpichai

Lots of love for Gemma 4! Team just told me it’s already had 10M+ downloads since last week’s launch. Gemma models have now been downloaded 500M+ times! Excited to see what you all are creating 👀

💬20🔄16❤️454👀40853📊49

As many observed, that was not the right YouTube link. Here's the correct one. https://t.co/Vui7wY...

Jeff Dean (@JeffDean) — Wed, 08 Apr 2026 23:10:32 GMT

As many observed, that was not the right YouTube link.  Here's the correct one.

youtube.com/watch?v=KKbgul…

💬1🔄4❤️68👀19791📊11

Hedged requests (apparently inspired by the Tail at Scale paper by myself and Luiz Barroso) applied ...

Jeff Dean (@JeffDean) — Wed, 08 Apr 2026 15:56:52 GMT

Hedged requests (apparently inspired by the Tail at Scale paper by myself and Luiz Barroso) applied within a single machine to replicating data across DRAM channels and issuing reads to all channels, using the one that comes back first.  ~5-15X reduction in p99.99 read latency.

github.com/LaurieWired/ta…

Cool stuff, @lauriewired!

Accompanying video forwarded to me by a friend, which is how I learned about it:
youtube.com/watch?v=QFi2WV…

💬23🔄85❤️1308👀108476📊234

Have Too Many Tabs Open? G O V E R T I C A L !

Jeff Dean (@JeffDean) — Tue, 07 Apr 2026 17:22:40 GMT

Have
Too
Many
Tabs
Open?

G
O

V
E
R
T
I
C
A
L
!

Google@Google

Too many @GoogleChrome tabs open? Try vertical tabs, rolling out now. 

Just right-click any Chrome window and select “Show Tabs Vertically” to move your tabs to the side of the browser window, making it easier to read page titles and manage tab groups.

Your browser does not support the video tag.

💬102🔄59❤️1374👀313623📊195

Incredible! The purple-ish and reddish colors don't quite come through when we look from Earth (and...

Jeff Dean (@JeffDean) — Tue, 07 Apr 2026 03:33:47 GMT

Incredible!  The purple-ish and reddish colors don't quite come through when we look from Earth (and the  black background highlights the vastness of space)

Physics & Astronomy Zone@zone_astronomy

The highest quality video of the moon was just released… this is so beautiful.

Your browser does not support the video tag.

💬48🔄111❤️2062👀253820📊227

How does a major news organization push out such an egregious headline error? 🤦‍♂️

Jeff Dean (@JeffDean) — Sat, 04 Apr 2026 16:12:48 GMT

How does a major news organization push out such an egregious headline error? 🤦‍♂️

Ben Phillips@benphillips76

Oh dear. 

The New York Times clever headline about NATO is entirely based on them getting its name wrong 😐

The A doesn’t stand for America, dudes.

💬158🔄52❤️2033👀302640📊259