正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

Introducing AI Browser Games! Watch open & closed models build small browser games head to head. Open models like Kimi K2.7 were faster, cheaper, & produced games similar to Opus 4.8. In some cases, open models were ~20x cheaper than closed ones too!

Hassan

99,520 subscribers

14,192 次观看 • 2 个月前 •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

18 条评论

Hassan 的头像

Hassan2 个月前

Try AI Browser games here! All powered by @togethercompute.

Steve Krouse 的头像

Steve Krouse2 个月前

sick!! also - huh?

Hassan 的头像

Hassan2 个月前

thanks! the card on top shows M3's avg cost for all runs and the row in the table shows M3's cost specifically for the breakout game. can probably make it more clear that the table rows are for each individual game and not the aggregate cost!

AL Robin 的头像

AL Robin2 个月前

I think it would be possible to achieve similar results completely free of charge, of course

Harshit Khemani 的头像

Harshit Khemani2 个月前

this ABG is doing numbers:

Hassan 的头像

Hassan2 个月前

whats ABG?

Hassan 的头像

Hassan2 个月前

omg duh 😂

Harshit Khemani 的头像

Harshit Khemani2 个月前

hahah

John 的头像

John2 个月前

20x cheaper is wild but the real question is consistency across runs. open models tend to have higher variance on creative tasks so the avg looks close but the floor is much lower

Ali A Ali 的头像

Ali A Ali2 个月前

It's crazy how quickly and cheaply these models build games that used to take large teams. Makes me wonder what it will take for AI to invent the games themselves. What's the token cost for creativity?

SupermanSpace 𝕏 的头像

SupermanSpace 𝕏2 个月前

This is similar:

Martin Ronfort 的头像

Martin Ronfort2 个月前

This is what happens when one path removes the efficiency constraint. Kimi building for limited hardware forced better engineering. Now they're outrunning Opus.

Fadi Hares 的头像

Fadi Hares2 个月前

The 20x cost gap is the headline everyone will miss. Speed and price parity means the moat isn't raw capability anymore, it's tooling and reliability. Curious how the open models held up on longer game logic, not just the first render?

Hussain Fakhruddin 的头像

Hussain Fakhruddin2 个月前

Open models matching Opus quality here is notable.

Gregor 的头像

Gregor2 个月前

20x cheaper settled it before the games even rendered

MaatWork 的头像

MaatWork2 个月前

Open models make redundant validators cheap. You shift from one-shot to consensus shipping (catching failures single passes miss). That changes how you build stuff.

Nick Venturi 的头像

Nick Venturi2 个月前

paying twenty times more for the same pixels makes zero sense

Curious Nori | AI · Food · Travel 的头像

Curious Nori | AI · Food · Travel2 个月前

Cool AI games 🎮

相关视频

$NVDA CEO Jensen Huang says Wall Street misunderstood DeepSeek and is “misunderstanding Kimi again.” He argues open Chinese models shouldn't be banned because cheaper open models accelerate AI adoption making “free AI” bullish for chips, data centers & Nvidia’s compute demand.

$NVDA CEO Jensen Huang says Wall Street misunderstood DeepSeek and is “misunderstanding Kimi again.” He argues open Chinese models shouldn't be banned because cheaper open models accelerate AI adoption making “free AI” bullish for chips, data centers & Nvidia’s compute demand.

Shay Boloor

187,570 次观看 • 1 个月前

"Closed models are cheaper. The reason you need open models is that they give you control". - Nvidia CEO Jensen Huang "One of the things that people misunderstand about open models is this: yes, you can host them yourself, and you can build your own computer. But most people use computers in the cloud. Frankly, I think closed models are cheaper. If you do not build everything yourself, you avoid the significant cost and expertise required to train, fine-tune, maintain, guardrail, evaluate, and keep the model safe. You would also need to build or obtain the computers required to host it. So, there is nothing cheap about doing all of that yourself. The reason you need open models is that they give you control. You can adapt them for your own highly specialized use cases. I think there is a lot of misunderstanding about closed models versus open models. We felt it was important for people to understand that there is a place for both." ---- From "Bloomberg Podcasts" YouTube channel, (full video link in comment)

"Closed models are cheaper. The reason you need open models is that they give you control". - Nvidia CEO Jensen Huang "One of the things that people misunderstand about open models is this: yes, you can host them yourself, and you can build your own computer. But most people use computers in the cloud. Frankly, I think closed models are cheaper. If you do not build everything yourself, you avoid the significant cost and expertise required to train, fine-tune, maintain, guardrail, evaluate, and keep the model safe. You would also need to build or obtain the computers required to host it. So, there is nothing cheap about doing all of that yourself. The reason you need open models is that they give you control. You can adapt them for your own highly specialized use cases. I think there is a lot of misunderstanding about closed models versus open models. We felt it was important for people to understand that there is a place for both." ---- From "Bloomberg Podcasts" YouTube channel, (full video link in comment)

Rohan Paul

32,317 次观看 • 3 天前

GLM 5.2 by Z.ai is even bigger than the DeepSeek moment. In the past few months, open source AI had been quite stagnant. New releases were underwhelming and still far behind the top closed models. But GLM 5.2 is different. It gets surprisingly close to GPT and Opus-level quality, while being much cheaper and open-weight. It just gets things done with very minimal hand holding. Full review

GLM 5.2 by Z.ai is even bigger than the DeepSeek moment. In the past few months, open source AI had been quite stagnant. New releases were underwhelming and still far behind the top closed models. But GLM 5.2 is different. It gets surprisingly close to GPT and Opus-level quality, while being much cheaper and open-weight. It just gets things done with very minimal hand holding. Full review

AI Search

36,494 次观看 • 2 个月前

Today we announced the Kaggle Game Arena, a new benchmarking platform where AI models and agents can compete head-to-head in strategic games, starting with chess ♟️. Why games, you ask? 🤔 Games are perfect for AI evaluation because they help us understand how models tackle complex reasoning tasks. Many games are a proxy for real-world skills and can test a model's ability in areas like strategic planning, adaptation, and memory. To inaugurate the Game Arena, we're launching an AI chess exhibition tournament. The world’s leading AI models will battle in head-to-head games in a multi-day event running from August 5-7 with streamed games happening daily. Tune-in to root for your favorites at

Today we announced the Kaggle Game Arena, a new benchmarking platform where AI models and agents can compete head-to-head in strategic games, starting with chess ♟️. Why games, you ask? 🤔 Games are perfect for AI evaluation because they help us understand how models tackle complex reasoning tasks. Many games are a proxy for real-world skills and can test a model's ability in areas like strategic planning, adaptation, and memory. To inaugurate the Game Arena, we're launching an AI chess exhibition tournament. The world’s leading AI models will battle in head-to-head games in a multi-day event running from August 5-7 with streamed games happening daily. Tune-in to root for your favorites at

Google AI

132,302 次观看 • 1 年前

🤯KIMI K3 ABSOLUTELY MOGS! BEATING Opus 4.8, GPT 5.5, and even Fable 5 in multiple benchmarks. They scored 1688 on GDPval-AA v2 🔥 This is a completely different breed of open-source models Kimi creates better games and front-end designs than Fable 5, but it's 8x cheaper! The results coming out are truly impressive! I will be testing this out further and posting multiple tests today. Stay tuned!

🤯KIMI K3 ABSOLUTELY MOGS! BEATING Opus 4.8, GPT 5.5, and even Fable 5 in multiple benchmarks. They scored 1688 on GDPval-AA v2 🔥 This is a completely different breed of open-source models Kimi creates better games and front-end designs than Fable 5, but it's 8x cheaper! The results coming out are truly impressive! I will be testing this out further and posting multiple tests today. Stay tuned!

Mark Santos

128,094 次观看 • 1 个月前

Claude Fable 5 vs Qwen3.8-Max i build Mario game with the same single prompt in both models qwen was 7.5x cheaper than fable yet it competes with one of the strongest AI model is China finally leading the AI race in open models?

Claude Fable 5 vs Qwen3.8-Max i build Mario game with the same single prompt in both models qwen was 7.5x cheaper than fable yet it competes with one of the strongest AI model is China finally leading the AI race in open models?

Tim Jayas

10,433 次观看 • 1 个月前

Stop using browser agents. You can use your coding agent to generate deterministic workflows that are 80% faster than browser-use and cost $0 in tokens Completely free & open source at

Stop using browser agents. You can use your coding agent to generate deterministic workflows that are 80% faster than browser-use and cost $0 in tokens Completely free & open source at

Tanishq (tk)

53,052 次观看 • 2 个月前

Demis says he wants to see a Western open source AI stack and that we’re losing to China. He also says Google doesn’t have enough compute to build two frontier (open and closed) models, which is why Gemma is a smaller family of models. Watch this incredible clip. Shout out Y Combinator and Garry Tan for the fantastic interview.

Demis says he wants to see a Western open source AI stack and that we’re losing to China. He also says Google doesn’t have enough compute to build two frontier (open and closed) models, which is why Gemma is a smaller family of models. Watch this incredible clip. Shout out Y Combinator and Garry Tan for the fantastic interview.

Matthew Berman

296,693 次观看 • 4 个月前

Kevin explains how frontier AI labs could benefit from open-sourcing models. “If they were to verticalize more quickly... there’s a tipping point where open-sourcing some of their models can make sense because they have a way to potentially capture more value.”

Kevin explains how frontier AI labs could benefit from open-sourcing models. “If they were to verticalize more quickly... there’s a tipping point where open-sourcing some of their models can make sense because they have a way to potentially capture more value.”

Delphi Digital

28,682 次观看 • 1 个月前

BREAKING: Max Cook (Max Cook), Sector Head at Coatue (COATUE) says.. "The keyboard & mouse are slowly dying." :( Cook also offers a bigger Q for the market: "What happens when OpenAI or Anthropic decide to offer smaller, cheaper, fine-tuned production models as products?" Cook argues the real AI debate might actually not be open vs. closed models It's what happens when frontier labs begin competing across the entire model stack.. Offering everything from cutting-edge reasoning models to smaller, efficient, low-cost, production-grade models for enterprise

BREAKING: Max Cook (Max Cook), Sector Head at Coatue (COATUE) says.. "The keyboard & mouse are slowly dying." :( Cook also offers a bigger Q for the market: "What happens when OpenAI or Anthropic decide to offer smaller, cheaper, fine-tuned production models as products?" Cook argues the real AI debate might actually not be open vs. closed models It's what happens when frontier labs begin competing across the entire model stack.. Offering everything from cutting-edge reasoning models to smaller, efficient, low-cost, production-grade models for enterprise

Molly O’Shea

304,663 次观看 • 2 个月前

Introducing NVFP4 — a new, AI-optimized data format that makes it faster, cheaper and greener to train AI models that can be applied to fields like healthcare, transportation and energy while preserving accuracy. Watch as Dion Harris explains our latest NVIDIA innovation:

Introducing NVFP4 — a new, AI-optimized data format that makes it faster, cheaper and greener to train AI models that can be applied to fields like healthcare, transportation and energy while preserving accuracy. Watch as Dion Harris explains our latest NVIDIA innovation:

NVIDIA Newsroom

49,004 次观看 • 1 年前

the market has evolved and Kimi K3 is bullish for the infrastructure layer "There’ll be an interesting, almost thermonuclear battle to provide the compute to support the demand that’s emerging for open weight models" -Benchmark's Peter Fenton Deepseek 1.5 yrs ago made everyone question how much compute AI would actually need. K3 points the other way -- cheaper, open models mean wider distribution, more usage, more inference

the market has evolved and Kimi K3 is bullish for the infrastructure layer "There’ll be an interesting, almost thermonuclear battle to provide the compute to support the demand that’s emerging for open weight models" -Benchmark's Peter Fenton Deepseek 1.5 yrs ago made everyone question how much compute AI would actually need. K3 points the other way -- cheaper, open models mean wider distribution, more usage, more inference

Deirdre Bosa

38,978 次观看 • 1 个月前

Before Claude Fable 5 got banned, I turned all my fine-tuning research and experiments into a product: (A CLI for data generation to fine-tune AI models). Finetuner uses Codex 5.5 or Opus 4.8 as the orchestrator and Chinese models (DeepSeek v4 Pro, Kimi K2.7, MiMo v2.5, etc.) to generate dataset rows. This system is unique because these are not Python-script-paraphrased datasets. Each and every row is handcrafted by Chinese models based on the orchestrator model's instructions. Now anyone can generate datasets to fine-tune small language models (1B to 30B models). I achieved 10x lower costs and 5x better dataset quality. Releasing the product in a few days. (I'll open-source the skills.)

Before Claude Fable 5 got banned, I turned all my fine-tuning research and experiments into a product: (A CLI for data generation to fine-tune AI models). Finetuner uses Codex 5.5 or Opus 4.8 as the orchestrator and Chinese models (DeepSeek v4 Pro, Kimi K2.7, MiMo v2.5, etc.) to generate dataset rows. This system is unique because these are not Python-script-paraphrased datasets. Each and every row is handcrafted by Chinese models based on the orchestrator model's instructions. Now anyone can generate datasets to fine-tune small language models (1B to 30B models). I achieved 10x lower costs and 5x better dataset quality. Releasing the product in a few days. (I'll open-source the skills.)

CJ Zafir

40,211 次观看 • 2 个月前

Genspark has released a new AI browser that gives you access to: - 169 open source models that run LOCALLY - Automation of browsing and workflows - Agents on any page And much more. Available on Mac and Windows (use cases below)

Genspark has released a new AI browser that gives you access to: - 169 open source models that run LOCALLY - Automation of browsing and workflows - Agents on any page And much more. Available on Mac and Windows (use cases below)

Paul Couvert

80,955 次观看 • 11 个月前

🔧🤖 New wave of open-source LLMs like Deekseek-R1-0528 and Qwen3-235B-A22B are leveling up with stronger agentic performance. We test them in head-to-head gameplay — the upgraded Deekseek-R1-0528 outsmarts strong reasoning models like o4-mini across several games and it nearly matches SOTA performance on Tetris, going toe-to-toe with o3. ✨🧠 Check out how R1 manages to clear lines in Tetris while other models still struggle 👇

🔧🤖 New wave of open-source LLMs like Deekseek-R1-0528 and Qwen3-235B-A22B are leveling up with stronger agentic performance. We test them in head-to-head gameplay — the upgraded Deekseek-R1-0528 outsmarts strong reasoning models like o4-mini across several games and it nearly matches SOTA performance on Tetris, going toe-to-toe with o3. ✨🧠 Check out how R1 manages to clear lines in Tetris while other models still struggle 👇

Hao AI Lab

36,573 次观看 • 1 年前

The open-vs.-closed AI debate may be asking the wrong question. NVIDIA’s Director of Develop Tech Nader Khalil (Nader Khalil🍊) joined Patrick Moorhead at The Six Five Summit: AI Unleashed to explain why open source doesn’t need to “win” to reshape AI. Open models can accelerate innovation across both open and proprietary ecosystems. Watch their AI Platforms spotlight conversation now on demand:

The open-vs.-closed AI debate may be asking the wrong question. NVIDIA’s Director of Develop Tech Nader Khalil (Nader Khalil🍊) joined Patrick Moorhead at The Six Five Summit: AI Unleashed to explain why open source doesn’t need to “win” to reshape AI. Open models can accelerate innovation across both open and proprietary ecosystems. Watch their AI Platforms spotlight conversation now on demand:

Six Five Media

36,866 次观看 • 10 天前

4 years ago we were on the brink of AI becoming proprietary and centralized, when OpenAI kept GPT3 closed and VCs started dumping money on researchers. From fully open science, to fully closed, in a matter of months. It was scary, and 1,000+ leading researchers and scientists banded together to show the world that it was possible to do the same work in the open, and build an ecosystem that benefits everyone. That was the BigScience Research Workshop BLOOM project, and it put us back on track to open science, starting with forward-thinking organizations like Meta releasing OPT. Look at us now. Open models have not only caught up, they're state of the art now. Not just LLMs, but models for document AI, speech to text, text to speech, generating images and more. We're closing in on 2 million open weight models on Hugging Face. Thanks for the reminder Thomas Wolf .

4 years ago we were on the brink of AI becoming proprietary and centralized, when OpenAI kept GPT3 closed and VCs started dumping money on researchers. From fully open science, to fully closed, in a matter of months. It was scary, and 1,000+ leading researchers and scientists banded together to show the world that it was possible to do the same work in the open, and build an ecosystem that benefits everyone. That was the BigScience Research Workshop BLOOM project, and it put us back on track to open science, starting with forward-thinking organizations like Meta releasing OPT. Look at us now. Open models have not only caught up, they're state of the art now. Not just LLMs, but models for document AI, speech to text, text to speech, generating images and more. We're closing in on 2 million open weight models on Hugging Face. Thanks for the reminder Thomas Wolf .

Jeff Boudier 🤗

21,947 次观看 • 1 年前

📢Introducing Kaggle Game Arena: a new, open benchmark platform where top AI models compete in complex, strategic games in streamed match-ups. We're charting new frontiers for trustworthy AI evaluation and it begins with chess — a classic proving ground for system intelligence.

📢Introducing Kaggle Game Arena: a new, open benchmark platform where top AI models compete in complex, strategic games in streamed match-ups. We're charting new frontiers for trustworthy AI evaluation and it begins with chess — a classic proving ground for system intelligence.

Kaggle

114,176 次观看 • 1 年前

📡 Transmission detected. Brave Games kicks off tomorrow with Game 1. Thousands are already in, but it’s not too late to join. Registration stays open throughout the Games. Download Brave, pick your faction, and join the heist. Your browser is the playing field now. 🕶️↓

📡 Transmission detected. Brave Games kicks off tomorrow with Game 1. Thousands are already in, but it’s not too late to join. Registration stays open throughout the Games. Download Brave, pick your faction, and join the heist. Your browser is the playing field now. 🕶️↓

Basic Attention Token (BAT)

34,713 次观看 • 7 个月前