Video wird geladen...

Video konnte nicht geladen werden

Zur Startseite

What if pathology foundation models could do more with less? GigaPath-Flash and GigaTIME-Flash cut computational demands while maintaining strong performance, opening the door to larger studies and broader exploration.

MSFTResearch's profile picture

Microsoft Research

554,893 subscribers

18,174 Aufrufe • vor 15 Tagen •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

2 Kommentare

Profilbild von Fiducial.ai
Fiducial.aivor 14 Tagen

Population-scale runs need a versioned feature manifest for the encoder, tiling and scanner preprocessing. The compute saving loses value if cohorts silently mix incompatible representations.

Profilbild von Neeraj Kumar
Neeraj Kumarvor 15 Tagen

How do the Flash models hold up on external cohorts and rare morphology classes? Compute savings matter only if calibration survives scanner and site shift.

Ähnliche Videos

🚨 You can use GPT-6 Astra as the orchestrator and DeepSeek V4.1 Flash as the agents to cut costs by more than 88% while maintaining the same level of quality. And now it’s even better, since DeepSeek V4.1 Flash is one of the best models for cybersecurity at an incredibly low price
0:40

🚨 You can use GPT-6 Astra as the orchestrator and DeepSeek V4.1 Flash as the agents to cut costs by more than 88% while maintaining the same level of quality. And now it’s even better, since DeepSeek V4.1 Flash is one of the best models for cybersecurity at an incredibly low price

Iam_

32,334 Aufrufe • vor 6 Tagen

introducing Gemini 3.6 Flash: our latest model that balances speed with intelligence to deliver strong performance in agentic and multimodal tasks built directly on developer and customer feedback from 3.5 Flash available now in AI Studio and the Gemini API
0:58

introducing Gemini 3.6 Flash: our latest model that balances speed with intelligence to deliver strong performance in agentic and multimodal tasks built directly on developer and customer feedback from 3.5 Flash available now in AI Studio and the Gemini API

Google AI Studio

134,158 Aufrufe • vor 1 Monat

Our new Gemini 3.5 Flash is our strongest agentic and coding model yet, delivering frontier-level performance at 4x the speed of comparable frontier models at less than half the cost. We had Tulsee Doshi break down what you need to know about it.
1:24

Our new Gemini 3.5 Flash is our strongest agentic and coding model yet, delivering frontier-level performance at 4x the speed of comparable frontier models at less than half the cost. We had Tulsee Doshi break down what you need to know about it.

Google

77,764 Aufrufe • vor 3 Monaten

Blank Bio (Blank Bio) is building foundation models to power a computational toolkit for RNA therapeutics, starting with mRNA design and expanding to target ID, biomarker discovery, and more. Congrats on the launch, Jonny, Phil Fradkin & Ian Shi!
2:56

Blank Bio (Blank Bio) is building foundation models to power a computational toolkit for RNA therapeutics, starting with mRNA design and expanding to target ID, biomarker discovery, and more. Congrats on the launch, Jonny, Phil Fradkin & Ian Shi!

Y Combinator

96,901 Aufrufe • vor 1 Jahr

What if we could RECREATE the REFLECTION of the FLASH seen in the GLASS during the Charlie Kirk assassination? Literally recreate it on site. The timing of the flash seen in the reflection of the glass is what makes it so interesting. What else would cause a flash that bright in the middle of the day? This could at least determine a lot of good information and confirm or deny theories that are out there. Just to get an idea of where the FLASH came from, which could identify a possible shooter or 2nd shooter location. The reason I say this is because it lines up perfectly with the sound of the gun, meaning you see the flash first, then you hear the shot, which would make sense because light travels faster than sound. This way we could get an idea of where it came from in the first place and determine if it's a good lead or not and then check the crowd or building locations. It could be recreated with a camera flash or something with bright enough lumens. This could clear up at least one theory or confirm it and lead us in the right direction of where the shooter could have been. Please comment with your thoughts and ideas. We need to get someone out there to the location with the Proper equipment. At least we could confirm or deny some theories. Video in bottom left shows the flash.
0:52

What if we could RECREATE the REFLECTION of the FLASH seen in the GLASS during the Charlie Kirk assassination? Literally recreate it on site. The timing of the flash seen in the reflection of the glass is what makes it so interesting. What else would cause a flash that bright in the middle of the day? This could at least determine a lot of good information and confirm or deny theories that are out there. Just to get an idea of where the FLASH came from, which could identify a possible shooter or 2nd shooter location. The reason I say this is because it lines up perfectly with the sound of the gun, meaning you see the flash first, then you hear the shot, which would make sense because light travels faster than sound. This way we could get an idea of where it came from in the first place and determine if it's a good lead or not and then check the crowd or building locations. It could be recreated with a camera flash or something with bright enough lumens. This could clear up at least one theory or confirm it and lead us in the right direction of where the shooter could have been. Please comment with your thoughts and ideas. We need to get someone out there to the location with the Proper equipment. At least we could confirm or deny some theories. Video in bottom left shows the flash.

The SCIF

108,760 Aufrufe • vor 11 Monaten

⚡️🔬📣 Excited to share our new nature article building and evaluating PathChat, a multimodal generative AI copilot and chatbot for human pathology. Article: Open Access Link: We leverage our previous success in building foundation models for computational pathology such as UNI / CONCH and combine it with the advancements of large vision language models and generative AI to enable PathChat to answer diverse pathology-related queries. We assessed PathChat using both multiple choice diagnostic questions and open-ended questions. Congratulations to Max Lu Bowen Chen @DFKW_MD Richard J. Chen and everyone else who contributed to this work. Also see blog post from Max Lu about this work: , also teasing the development and preview of PathChat 2, a successor to PathChat 1 bringing new capabilities and substantially improved performance to the state-of-the-art.
1:30

⚡️🔬📣 Excited to share our new nature article building and evaluating PathChat, a multimodal generative AI copilot and chatbot for human pathology. Article: Open Access Link: We leverage our previous success in building foundation models for computational pathology such as UNI / CONCH and combine it with the advancements of large vision language models and generative AI to enable PathChat to answer diverse pathology-related queries. We assessed PathChat using both multiple choice diagnostic questions and open-ended questions. Congratulations to Max Lu Bowen Chen @DFKW_MD Richard J. Chen and everyone else who contributed to this work. Also see blog post from Max Lu about this work: , also teasing the development and preview of PathChat 2, a successor to PathChat 1 bringing new capabilities and substantially improved performance to the state-of-the-art.

Faisal Mahmood

291,625 Aufrufe • vor 2 Jahren

Aligned with Matan Grinberg's intuition that in the next 12 months, 90% of tokens could be going to open models. We're seeing more and more companies using frontier APIs for experimenting and open-source or private models for production and larger workloads!
1:08

Aligned with Matan Grinberg's intuition that in the next 12 months, 90% of tokens could be going to open models. We're seeing more and more companies using frontier APIs for experimenting and open-source or private models for production and larger workloads!

clem 🤗

24,138 Aufrufe • vor 2 Monaten

🏠 Everyone talks about building more homes. Almost no one talks about whether federal mortgage policy is adding fuel to the fire. If more government-backed financing creates more buyers while housing remains limited, what happens to prices? Allen Mendenhall is a Research Fellow in Economic Policy Studies and Senior Advisor for the Free Enterprise Initiative at The Heritage Foundation. John Gibbs is the Director of Economic Policy Studies at the Heritage Foundation.
1:24

🏠 Everyone talks about building more homes. Almost no one talks about whether federal mortgage policy is adding fuel to the fire. If more government-backed financing creates more buyers while housing remains limited, what happens to prices? Allen Mendenhall is a Research Fellow in Economic Policy Studies and Senior Advisor for the Free Enterprise Initiative at The Heritage Foundation. John Gibbs is the Director of Economic Policy Studies at the Heritage Foundation.

Heritage Foundation

14,964 Aufrufe • vor 1 Monat

Introducing Gemini 3.5 Flash ⚡️ Normally, Gemini 3.5 Flash is 4x faster than other models with frontier performance. For a limited time, Antigravity is serving it 12x faster thanks to custom inference tricks, delivering incredible speed for your workflows. 🚀 See the performance in this demo: generating pixel art from photos, orchestrating multi-agent workflows to write and register sprites, and spawning browser subagents to auto-test rendering: 👇
1:08

Introducing Gemini 3.5 Flash ⚡️ Normally, Gemini 3.5 Flash is 4x faster than other models with frontier performance. For a limited time, Antigravity is serving it 12x faster thanks to custom inference tricks, delivering incredible speed for your workflows. 🚀 See the performance in this demo: generating pixel art from photos, orchestrating multi-agent workflows to write and register sprites, and spawning browser subagents to auto-test rendering: 👇

Google Antigravity

132,207 Aufrufe • vor 3 Monaten

Our most intelligent workhorse model yet for coding and agents has arrived ⚡ Meet Gemini 3.7 Flash. — Crush that seemingly endless to-do list. Gemini Spark in the Google Gemini now uses 3.7 Flash. The new model can equip your personal AI agent to work even smarter for you by seamlessly handling complex, multi-step tasks across your Google Workspace apps like Gmail, Google Calendar and Google Docs — Enjoy a smoother build experience. The model thinks more diligently, putting more effort into multi-step planning and tool calls. A more disciplined execution means less manual oversight and fewer retries across engineering workflows — Build more, spend less. 3.7 Flash is available through the end of the year with an introductory price of half the original 3.6 Flash cost per million tokens ($0.75/1M input tokens and $3.75/1M output tokens)
1:15

Our most intelligent workhorse model yet for coding and agents has arrived ⚡ Meet Gemini 3.7 Flash. — Crush that seemingly endless to-do list. Gemini Spark in the Google Gemini now uses 3.7 Flash. The new model can equip your personal AI agent to work even smarter for you by seamlessly handling complex, multi-step tasks across your Google Workspace apps like Gmail, Google Calendar and Google Docs — Enjoy a smoother build experience. The model thinks more diligently, putting more effort into multi-step planning and tool calls. A more disciplined execution means less manual oversight and fewer retries across engineering workflows — Build more, spend less. 3.7 Flash is available through the end of the year with an introductory price of half the original 3.6 Flash cost per million tokens ($0.75/1M input tokens and $3.75/1M output tokens)

Google AI

128,730 Aufrufe • vor 1 Monat

Computing hardware architecture has evolved from maximizing single-core performance to multiplying parallel processing capacity through more cores, more threads, and higher computational density. This transition demands software architectures specifically designed to exploit parallel processing, especially in advanced AI and blockchain-based systems. As Greg Meredith, CEO of our partner F1R3FLY. io, notes, "We need software architectures that can eat physical threads of computation and turn that into throughput." Traditional sequential programming models often struggle to effectively leverage the potential of these multi-core systems, creating a bottleneck in performance scaling. Rholang addresses this challenge through its foundation in the Rho-Calculus, a reflective higher-order process calculus specifically designed for inherent concurrency and reflection. Unlike computational models based on sequential execution or those that bolt on concurrency primitives, Rholang can leverage its process-oriented nature to detect and distribute non-interfering logical computation threads across available physical processing units, creating a more direct correlation between hardware resources and performance scaling. The Rho-Calculus was specifically designed to provide the minimal set of operations needed to model autonomous agents operating in parallel while maintaining the ability to represent their own reasoning processes. This structure mirrors the operational characteristics of intelligent systems, where autonomous processes run independently while communicating and coordinating. For our novel decentralized AI platform, MeTTaCycle, this future-proof architectural foundation enables more than just enhanced transactional throughput; it fosters an environment where independent computational processes can efficiently coexist and interact, forming a critical layer for the Artificial Superintelligence Alliance's decentralized AGI infrastructure and products.
1:42

Computing hardware architecture has evolved from maximizing single-core performance to multiplying parallel processing capacity through more cores, more threads, and higher computational density. This transition demands software architectures specifically designed to exploit parallel processing, especially in advanced AI and blockchain-based systems. As Greg Meredith, CEO of our partner F1R3FLY. io, notes, "We need software architectures that can eat physical threads of computation and turn that into throughput." Traditional sequential programming models often struggle to effectively leverage the potential of these multi-core systems, creating a bottleneck in performance scaling. Rholang addresses this challenge through its foundation in the Rho-Calculus, a reflective higher-order process calculus specifically designed for inherent concurrency and reflection. Unlike computational models based on sequential execution or those that bolt on concurrency primitives, Rholang can leverage its process-oriented nature to detect and distribute non-interfering logical computation threads across available physical processing units, creating a more direct correlation between hardware resources and performance scaling. The Rho-Calculus was specifically designed to provide the minimal set of operations needed to model autonomous agents operating in parallel while maintaining the ability to represent their own reasoning processes. This structure mirrors the operational characteristics of intelligent systems, where autonomous processes run independently while communicating and coordinating. For our novel decentralized AI platform, MeTTaCycle, this future-proof architectural foundation enables more than just enhanced transactional throughput; it fosters an environment where independent computational processes can efficiently coexist and interact, forming a critical layer for the Artificial Superintelligence Alliance's decentralized AGI infrastructure and products.

SingularityNET

29,368 Aufrufe • vor 1 Jahr

deepseek v4.1 flash vs secret model tested both with the same prompt at the highest reasoning available > v4.1 flash took 2 hours to finish both tests and cost $2.60 > secret model finished both in 30 minutes, can’t reveal cost yet v4.1 flash spent a lot of time overthinking and running random tests, which slowed it down thanks AIHubmix for providing api access so i could test v4.1 flash hoping that team will cut down the overthinking before the full release, but for now here are the results which model did better here?
0:41

deepseek v4.1 flash vs secret model tested both with the same prompt at the highest reasoning available > v4.1 flash took 2 hours to finish both tests and cost $2.60 > secret model finished both in 30 minutes, can’t reveal cost yet v4.1 flash spent a lot of time overthinking and running random tests, which slowed it down thanks AIHubmix for providing api access so i could test v4.1 flash hoping that team will cut down the overthinking before the full release, but for now here are the results which model did better here?

J A Z I I

28,763 Aufrufe • vor 7 Tagen

Three years ago, Gemini started by understanding the world. With Gemini 2, models learned to think and reason. Late last year, Gemini 3 brought any idea to life. Today, we’re continuing that journey with our Gemini 3.5 series, starting with Gemini 3.5 Flash, delivering frontier performance for agents and coding.
1:01

Three years ago, Gemini started by understanding the world. With Gemini 2, models learned to think and reason. Late last year, Gemini 3 brought any idea to life. Today, we’re continuing that journey with our Gemini 3.5 series, starting with Gemini 3.5 Flash, delivering frontier performance for agents and coding.

Google AI

69,665 Aufrufe • vor 3 Monaten

The reason why people are sick usually has a lot less to do with physiology and a lot more to do with pathology in our politics. Air quality, water quality, whether you can walk in your community safely, whether or not you can see a doctor if you get sick in the first place.
0:54

The reason why people are sick usually has a lot less to do with physiology and a lot more to do with pathology in our politics. Air quality, water quality, whether you can walk in your community safely, whether or not you can see a doctor if you get sick in the first place.

Dr. Abdul El-Sayed

12,268 Aufrufe • vor 5 Monaten

glm 5.3 vs qwen 3.8 vs gemini 3.7 vs deepseek v4 flash four models designed and built three structures each on a physics-backed site, with no dimensions anywhere in the brief the setup: our own agent loop on OpenRouter, a construction site as the tool set – footings, walls, arches, roofs, scaffold, a lamp. the site enforces physics and nothing else: unsupported brick falls, a roof needs walls under it, a worker reaches 3.2 m above whatever he stands on, an arch needs centring until the keystone is set, concrete cures before it carries. no budget ceiling – material cost is tallied and reported, never blocked. tasks: 1. house – a plot and a palette, no plan. shape, height and material are the model's call 2. lighthouse – a headland cut by a gully, with a rock stack standing 30 m offshore. the lamp must burn, it must be the highest thing built, and the keeper must be able to walk to it 3. bridge – a river with one islet and banks at different heights. cross it however you want models: Z.ai glm 5.3 flash, Qwen qwen 3.8 flash, Google DeepMind gemini 3.7 flash, DeepSeek v4 flash vision all twelve objects were finished and signed off by the models themselves. tallest lighthouse is qwen's at 38.4 m, planted on the offshore stack with a bridge run out to it – the only model that read the site that way. deepseek signed off its bridge on an empty riverbed: 0 bricks, 107 minutes, $1.16m of material tallied - total cost, three builds #1 glm 5.3 flash – $0.201 #2 gemini 3.7 flash – $0.871 #3 qwen 3.8 flash – $1.058 #4 deepseek v4 flash – $1.567 - wall clock, three builds #1 gemini 3.7 flash – 91m #2 glm 5.3 flash – 228m #3 deepseek v4 flash – 502m #4 qwen 3.8 flash – 912m - total tokens #1 gemini 3.7 flash – 3,567,052 #2 glm 5.3 flash – 4,732,748 #3 qwen 3.8 flash – 13,469,333 #4 deepseek v4 flash – 18,230,076 - defects logged by the site #1 deepseek v4 flash – 59 #2 gemini 3.7 flash – 132 #3 glm 5.3 flash – 221 #4 qwen 3.8 flash – 350 - material tallied across three builds #1 gemini 3.7 flash – $359,884 #2 glm 5.3 flash – $583,358 #3 deepseek v4 flash – $1,327,484 #4 qwen 3.8 flash – $2,188,625 observations: • glm is the cheap one and nothing here is close – $0.201 for three buildings, $0.042 per million tokens, 6x under gemini's rate • what glm spends it on is bulk, not care: 166,228 bricks in one house and 156 defect weight, the worst single object in the set • gemini is the efficiency line – 91 minutes and 3.57m tokens for all three and an eighth of qwen's clock • gemini also builds the smallest of everything. its lighthouse is 22.5 m against qwen's 38.4, its house 6.9 m against 19.3 • qwen is the maximalist: 1.18m bricks, $2.19m of material, tallest on all three tasks, and 912 minutes – 15 hours – to get there conclusion: twelve finished objects for $3.80 all in, and a 7.8x price spread between the cheapest model and the priciest! follow thehype. for 24/7 ai news, analysis and breakdowns
1:09

glm 5.3 vs qwen 3.8 vs gemini 3.7 vs deepseek v4 flash four models designed and built three structures each on a physics-backed site, with no dimensions anywhere in the brief the setup: our own agent loop on OpenRouter, a construction site as the tool set – footings, walls, arches, roofs, scaffold, a lamp. the site enforces physics and nothing else: unsupported brick falls, a roof needs walls under it, a worker reaches 3.2 m above whatever he stands on, an arch needs centring until the keystone is set, concrete cures before it carries. no budget ceiling – material cost is tallied and reported, never blocked. tasks: 1. house – a plot and a palette, no plan. shape, height and material are the model's call 2. lighthouse – a headland cut by a gully, with a rock stack standing 30 m offshore. the lamp must burn, it must be the highest thing built, and the keeper must be able to walk to it 3. bridge – a river with one islet and banks at different heights. cross it however you want models: Z.ai glm 5.3 flash, Qwen qwen 3.8 flash, Google DeepMind gemini 3.7 flash, DeepSeek v4 flash vision all twelve objects were finished and signed off by the models themselves. tallest lighthouse is qwen's at 38.4 m, planted on the offshore stack with a bridge run out to it – the only model that read the site that way. deepseek signed off its bridge on an empty riverbed: 0 bricks, 107 minutes, $1.16m of material tallied - total cost, three builds #1 glm 5.3 flash – $0.201 #2 gemini 3.7 flash – $0.871 #3 qwen 3.8 flash – $1.058 #4 deepseek v4 flash – $1.567 - wall clock, three builds #1 gemini 3.7 flash – 91m #2 glm 5.3 flash – 228m #3 deepseek v4 flash – 502m #4 qwen 3.8 flash – 912m - total tokens #1 gemini 3.7 flash – 3,567,052 #2 glm 5.3 flash – 4,732,748 #3 qwen 3.8 flash – 13,469,333 #4 deepseek v4 flash – 18,230,076 - defects logged by the site #1 deepseek v4 flash – 59 #2 gemini 3.7 flash – 132 #3 glm 5.3 flash – 221 #4 qwen 3.8 flash – 350 - material tallied across three builds #1 gemini 3.7 flash – $359,884 #2 glm 5.3 flash – $583,358 #3 deepseek v4 flash – $1,327,484 #4 qwen 3.8 flash – $2,188,625 observations: • glm is the cheap one and nothing here is close – $0.201 for three buildings, $0.042 per million tokens, 6x under gemini's rate • what glm spends it on is bulk, not care: 166,228 bricks in one house and 156 defect weight, the worst single object in the set • gemini is the efficiency line – 91 minutes and 3.57m tokens for all three and an eighth of qwen's clock • gemini also builds the smallest of everything. its lighthouse is 22.5 m against qwen's 38.4, its house 6.9 m against 19.3 • qwen is the maximalist: 1.18m bricks, $2.19m of material, tallest on all three tasks, and 912 minutes – 15 hours – to get there conclusion: twelve finished objects for $3.80 all in, and a 7.8x price spread between the cheapest model and the priciest! follow thehype. for 24/7 ai news, analysis and breakdowns

thehype.

26,250 Aufrufe • vor 18 Tagen

If you ever want a crash course on what not to do during a Flash Flood. Just visit Houston, Texas. This was the scene off i45 and despite literally dozens of vehicles stalled out more continued to try leading to water rescues. #TXwx
1:35

If you ever want a crash course on what not to do during a Flash Flood. Just visit Houston, Texas. This was the scene off i45 and despite literally dozens of vehicles stalled out more continued to try leading to water rescues. #TXwx

Aaron Rigsby

1,332,479 Aufrufe • vor 2 Jahren

Meteorologist Evan Bentley from the NWS Storm Prediction Center joined us on stream today and broke down two questions we hear a lot: What do I do in a high-rise with no basement? And what if I’m under a tornado warning AND a flash flood warning at the same time?
1:31

Meteorologist Evan Bentley from the NWS Storm Prediction Center joined us on stream today and broke down two questions we hear a lot: What do I do in a high-rise with no basement? And what if I’m under a tornado warning AND a flash flood warning at the same time?

Ryan Hall, Y’all

51,789 Aufrufe • vor 4 Monaten