Video yükleniyor...

Video Yüklenemedi

Bu video yüklenirken bir sorun oluştu. Bu geçici bir ağ sorunundan kaynaklanıyor olabilir veya video kullanılamıyor olabilir.

Ana Sayfaya Dön

This is the most important chart in AI right now. Open Source is eclipsing Closed Source in total token usage, but frontier labs still capture most revenue. Watch my breakdown:

Matthew Berman

153,102 subscribers

24,102 görüntüleme • 2 ay önce •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

22 Yorum

Brian Sowards (he/they) 🌅🇺🇸 profil fotoğrafı

Brian Sowards (he/they) 🌅🇺🇸2 ay önce

we gotta connect you to Nebius and their Token Factory Matthew, they’ve built an AI native cloud, inference optimization, and host the latest open source models i use codex as my driver and my subagents are OSS now

Matthew Berman profil fotoğrafı

Matthew Berman2 ay önce

I'm open to meeting them!

Darren profil fotoğrafı

Darren2 ay önce

Open source is only just arriving at this capability level

OdeToJoy profil fotoğrafı

OdeToJoy2 ay önce

You are completely ignoring the fact that the gap is closing really fast. The cost of being a frontier model provider cannot be covered in the 2-3 months they can stay in the lead before some dirt cheap open source and open weight model or hybrid setup outperforms them.

Cyril Dieumegard 🇨🇭 profil fotoğrafı

Cyril Dieumegard 🇨🇭2 ay önce

I just added the amazing chart to Very useful and interesting!

Donkey Kong profil fotoğrafı

Donkey Kong2 ay önce

Open weights is exactly how AI becomes successful!!! Props to all devs pushing OSS and OWMs 🙏

Ankit raj profil fotoğrafı

Ankit raj2 ay önce

There is no revenue except for nvidia, not yet

Nyla Pearson profil fotoğrafı

Nyla Pearson2 ay önce

That’s because the open source is parasitic - duh.. mainstream llms are the blood boy of open source

Ahmed Ibrahim Hamdy profil fotoğrafı

Ahmed Ibrahim Hamdy2 ay önce

Thanks @MatthewBerman for talking about this FLIP and surfacing it. I believe 80% of token consumption in the future will be on device using OpenWeight models with the remaining 20% being served by Cloud/Frontier Models.

John Kennedy Peterson profil fotoğrafı

John Kennedy Peterson2 ay önce

They “thought” … their programs were “the market”.. then opensource bitch slapped them back to reality…

Douglas Karr profil fotoğrafı

Douglas Karr2 ay önce

A “token” is not a fixed unit of text. It is whatever a given model’s tokenizer decided to treat as one piece. Your analysis is treating them as an equal unit of measurement across models.

IA Latinoamérica profil fotoğrafı

IA Latinoamérica2 ay önce

And this was recorded before the announcement regarding Ox Alpha being an open Chinese model...

Volodymyr Pavlenko profil fotoğrafı

Volodymyr Pavlenko2 ay önce

open source winning on usage while frontier labs keep the revenue is basically the story of every commoditization cycle

Seddik na tizit profil fotoğrafı

Seddik na tizit2 ay önce

It's also showing that China is winning hard and US is losing ground

Damon Gardenhire profil fotoğrafı

Damon Gardenhire2 ay önce

but the open source models just started catching up, so doesn't this data simply reflect that the "good enough" trend has not overtaken the current dynamics yet? But won't it quite soon? After all, when companies like Microsoft are slashing their spend on Anthropic and shifting toward open source models, it seems like those "rumors from the heights" haven't filtered down yet.

Joe profil fotoğrafı

Joe1 ay önce

It’s mostly vibecoders that can’t afford tokens because they’re building app with no revenue.

Jimmy Freedom profil fotoğrafı

Jimmy Freedom1 ay önce

Scam Altman didn’t build OpenAI

John Kennedy Peterson profil fotoğrafı

John Kennedy Peterson2 ay önce

“Chinese models” delusional … do you know how most AI systems are developed? The Chinese state has not made any AI models… the biggest “Chinese“ model was developed by an American educated H1B Visa holder while living in the United States and working in the United States #KIMI of Moonshot ….

Haydon Ryan 🇦🇺🇺🇸 profil fotoğrafı

Haydon Ryan 🇦🇺🇺🇸2 ay önce

Great take. Nothing stopping people trying open and only if that fails use the sorta closed. Again depends on speed, and like you said market competition.

Gill profil fotoğrafı

Gill2 ay önce

Open source drives high token volume while closed labs take enterprise cash.

dmtn dmtnm profil fotoğrafı

dmtn dmtnm2 ay önce

nobody knows how much revenue frontier capture

Chidi Okoene profil fotoğrafı

Chidi Okoene2 ay önce

Privacy is the main reason open source is winning on the big stage

Benzer Videolar

Sam Altman on Open Source AI's Future⁣ ⁣ "Right now, what everybody wants is just the most capable frontier coding model they can have. The big frontier models, even if we made them open source, are hard to run." — #sama

Sam Altman on Open Source AI's Future⁣ ⁣ "Right now, what everybody wants is just the most capable frontier coding model they can have. The big frontier models, even if we made them open source, are hard to run." — #sama

AI Insights

19,607 görüntüleme • 4 ay önce

The world changed this week Governments started banning frontier models. Hardware has become unobtainable The single most important thing you can be doing right now is getting into local AI In this video I cover EVERYTHING local AI. Most important video you'll watch this week:

The world changed this week Governments started banning frontier models. Hardware has become unobtainable The single most important thing you can be doing right now is getting into local AI In this video I cover EVERYTHING local AI. Most important video you'll watch this week:

Alex Finn

197,870 görüntüleme • 4 ay önce

The open-source & open-weight frontier is 7 months behind the closed-model frontier. Let’s talk about it. Check out last week’s episode for the reasons why you might want to consider open-source & open-weight models. Welcome to the fourth episode of the Open Model Series with merve, Hugging Face AI/ML engineer & open-sourceress! We have been discussing open-source & open-weight AI: what it is, how it works, and how you can run them on your own compute. Please leave all your questions about open-source & open-weight AI in the comments! This is a super important topic that is novel to most people, and it can be challenging to grasp - so rest assured your questions are shared with many others!

The open-source & open-weight frontier is 7 months behind the closed-model frontier. Let’s talk about it. Check out last week’s episode for the reasons why you might want to consider open-source & open-weight models. Welcome to the fourth episode of the Open Model Series with merve, Hugging Face AI/ML engineer & open-sourceress! We have been discussing open-source & open-weight AI: what it is, how it works, and how you can run them on your own compute. Please leave all your questions about open-source & open-weight AI in the comments! This is a super important topic that is novel to most people, and it can be challenging to grasp - so rest assured your questions are shared with many others!

Harper Carroll

17,279 görüntüleme • 1 ay önce

Sandeep thinks powerful AI in a few hands is the wrong endgame. Sentient's building a token-incentivized open-source AI instead. Proof it works? It became the most visited reasoning framework globally, not just in crypto, but across all AI.

Sandeep thinks powerful AI in a few hands is the wrong endgame. Sentient's building a token-incentivized open-source AI instead. Proof it works? It became the most visited reasoning framework globally, not just in crypto, but across all AI.

The Rollup

23,739 görüntüleme • 9 ay önce

"This is just like the space race." - Doug Kelly America must invest in open source and closed source AI to win the future.

"This is just like the space race." - Doug Kelly America must invest in open source and closed source AI to win the future.

American Edge Project

215,647 görüntüleme • 1 yıl önce

Most thinking folks know in their souls how important open source software is, even if they are hazy on the facts. The issue is sometimes folks forget. This will be true for AI, we in the US just need to remember. Here is the story of the Grandpa of open source software.

Most thinking folks know in their souls how important open source software is, even if they are hazy on the facts. The issue is sometimes folks forget. This will be true for AI, we in the US just need to remember. Here is the story of the Grandpa of open source software.

Brian Roemmele

29,110 görüntüleme • 7 ay önce

Open-source AI is booming, massively impactful for progress, competition, transparency & orders of magnitude less dangerous than closed-source frontier AI. Time to support it to the max!

Open-source AI is booming, massively impactful for progress, competition, transparency & orders of magnitude less dangerous than closed-source frontier AI. Time to support it to the max!

clem 🤗

52,739 görüntüleme • 4 ay önce

Dario Amodei is right. Most Chinese open source models are benchmark optimized illusions. They ace public tests and fall apart the moment you leave the playground. Held back evals expose the truth fast. Open source is still the future of AI. No question.

Dario Amodei is right. Most Chinese open source models are benchmark optimized illusions. They ace public tests and fall apart the moment you leave the playground. Held back evals expose the truth fast. Open source is still the future of AI. No question.

VraserX e/acc

48,596 görüntüleme • 7 ay önce

In an AI world, effort is no longer the bottleneck—ideas are. Open-source AI enables anyone to have access to build anything they want without depending on the one-sided gates that closed-source offers. This is amplified by the open innovation that open-source development enables.

In an AI world, effort is no longer the bottleneck—ideas are. Open-source AI enables anyone to have access to build anything they want without depending on the one-sided gates that closed-source offers. This is amplified by the open innovation that open-source development enables.

Sentient

17,408 görüntüleme • 8 ay önce

DeepSeek-V3 is live in AkashChat. This is the most capable open-source AI model available today — directly rivaling the benchmark performance of GPT-4o and 3.5 Sonnet.

DeepSeek-V3 is live in AkashChat. This is the most capable open-source AI model available today — directly rivaling the benchmark performance of GPT-4o and 3.5 Sonnet.

Akash Network

47,437 görüntüleme • 1 yıl önce

Yann LeCun just said something at Davos that nobody is talking about. The man who built Meta's AI for 12 years. The godfather of deep learning. Said this on stage, in front of the world's most powerful people: "The best open source AI models right now come from China. Everybody in the research community is using Chinese models." Here's what makes this more uncomfortable. He said it in the same breath as warning that the US is making a catastrophic mistake. OpenAI — increasingly closed. Anthropic — never open. Google — more closed every year. Meta — changing direction. Meanwhile China's labs are publishing. Open sourcing. Sharing. And the global research community is downloading their models. LeCun's warning was blunt: "We are slowing down progress in the West because of this." The irony is brutal. America built the open internet. Linux runs the world. The entire software stack humanity depends on is open source. And now the country that invented openness is closing its most important technology. While China does the opposite. The next AI revolution won't be won by the country with the most money. It will be won by the country whose models researchers actually use. Right now that answer is changing.

Yann LeCun just said something at Davos that nobody is talking about. The man who built Meta's AI for 12 years. The godfather of deep learning. Said this on stage, in front of the world's most powerful people: "The best open source AI models right now come from China. Everybody in the research community is using Chinese models." Here's what makes this more uncomfortable. He said it in the same breath as warning that the US is making a catastrophic mistake. OpenAI — increasingly closed. Anthropic — never open. Google — more closed every year. Meta — changing direction. Meanwhile China's labs are publishing. Open sourcing. Sharing. And the global research community is downloading their models. LeCun's warning was blunt: "We are slowing down progress in the West because of this." The irony is brutal. America built the open internet. Linux runs the world. The entire software stack humanity depends on is open source. And now the country that invented openness is closing its most important technology. While China does the opposite. The next AI revolution won't be won by the country with the most money. It will be won by the country whose models researchers actually use. Right now that answer is changing.

Shruti

31,709 görüntüleme • 5 ay önce

ELON: CHINA HAS THE BEST OPEN SOURCE AI MODELS RIGHT NOW “The best open source models are generally from China, which is bizarre. I think the second best one, or maybe it’s better than second best, is Grok 2.5. The open source model is actually very good, and we will continue to open source our models.” Source: The All-In Podcast

ELON: CHINA HAS THE BEST OPEN SOURCE AI MODELS RIGHT NOW “The best open source models are generally from China, which is bizarre. I think the second best one, or maybe it’s better than second best, is Grok 2.5. The open source model is actually very good, and we will continue to open source our models.” Source: The All-In Podcast

Mario Nawfal

1,163,582 görüntüleme • 1 yıl önce

Thank you for agreeing with us, Elon Musk. Open-source AI is better than AI built behind closed doors. We are optimistic about open source together with you in the future.

Thank you for agreeing with us, Elon Musk. Open-source AI is better than AI built behind closed doors. We are optimistic about open source together with you in the future.

Optimistic General Intelligence

62,397 görüntüleme • 2 ay önce

We still remember you! Happy anniversary to one of the most unique action games in history and our most important source of influence. Proud to carry the torch 🔥 What game is this? Check the video!

We still remember you! Happy anniversary to one of the most unique action games in history and our most important source of influence. Proud to carry the torch 🔥 What game is this? Check the video!

SPINE — This is Gun Fu

36,405 görüntüleme • 8 ay önce

DeepSeek-R1 is available in AkashChat. R1 is currently the most capable open-source AI model, surpassing 3.5 Sonnet and o1 on several leading benchmarks.

DeepSeek-R1 is available in AkashChat. R1 is currently the most capable open-source AI model, surpassing 3.5 Sonnet and o1 on several leading benchmarks.

Akash Network

21,184 görüntüleme • 1 yıl önce

This is not the moment for pacing. This is the moment for acceleration. I believe the push from the big AI labs to “slow down” isn’t about safety… it’s actually about regulatory capture. They’re watching the explosion in open source and they want regulators to save them. My take on Bloomberg: The big labs are not worried about recursive self-improvement. They are worried about open source. Token share for the frontier labs is dropping faster than anyone expected. Open source is rising faster than anyone expected. That is why NVIDIA just paid a record price for Hugging Face. Coming from healthcare, I know what regulatory capture looks like, and this looks and smells like regulatory capture. At the same time, the scaling laws are not paying off the way the big labs promised. Verticalized, sector specific small models with the right harnesses are starting to beat general purpose giants on the tasks that matter. That is very good news for healthcare, defense, biology, and every other domain where general intelligence was never really the point. So my POV on the Xi and Trump meeting later this month is simple. Put the guardrails on where AI is allowed to touch (defense, biology, gain of function, clinical decisions), not on who is allowed to build. Slowing US development while China goes all in on open source is not safety. It is a gift.

This is not the moment for pacing. This is the moment for acceleration. I believe the push from the big AI labs to “slow down” isn’t about safety… it’s actually about regulatory capture. They’re watching the explosion in open source and they want regulators to save them. My take on Bloomberg: The big labs are not worried about recursive self-improvement. They are worried about open source. Token share for the frontier labs is dropping faster than anyone expected. Open source is rising faster than anyone expected. That is why NVIDIA just paid a record price for Hugging Face. Coming from healthcare, I know what regulatory capture looks like, and this looks and smells like regulatory capture. At the same time, the scaling laws are not paying off the way the big labs promised. Verticalized, sector specific small models with the right harnesses are starting to beat general purpose giants on the tasks that matter. That is very good news for healthcare, defense, biology, and every other domain where general intelligence was never really the point. So my POV on the Xi and Trump meeting later this month is simple. Put the guardrails on where AI is allowed to touch (defense, biology, gain of function, clinical decisions), not on who is allowed to build. Slowing US development while China goes all in on open source is not safety. It is a gift.

Dr Danish

13,867 görüntüleme • 1 ay önce