Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

Muse Ember (Max) from AI at Meta just landed in the Code Arena: WebDev as the highest-performing small model - built for fast, high-quality, low-cost work. Watch the video below: 20+ websites and animations, all generated by Ember in one go. Across the generations we timed, the average was... show more

Arena.ai

240,024 subscribers

51,626 views • 7 hours ago •via X (Twitter)

Science & Technology

Anya Rossi• Live Now

Private livecam show

2 Comments

Arena.ai7 hours ago

Dig into the leaderboard details directly at:

Artzy7 hours ago

@AIatMeta That generation speed on multi site runs is wild

Related Videos

Agent Categories and Costs are now in the Agent Leaderboard! Two questions guide model selection: task-specific performance and cost to completion. Based on analysis of 1.7M+ real-world Agent Mode sessions, the leaderboard now lets users: - Filter the leaderboard by Code, Work, and Chat to compare model performance by task type - View real cost-per-task for each model, including the Pareto frontier of the most cost-efficient options at each performance level

Agent Categories and Costs are now in the Agent Leaderboard! Two questions guide model selection: task-specific performance and cost to completion. Based on analysis of 1.7M+ real-world Agent Mode sessions, the leaderboard now lets users: - Filter the leaderboard by Code, Work, and Chat to compare model performance by task type - View real cost-per-task for each model, including the Pareto frontier of the most cost-efficient options at each performance level

Arena.ai

29,605 views • 2 months ago

The business model in the AI era is completely different from that of Web era. Traditional internet companies thrive on user quantity, making it a competition for MAU’s. This was because the marginal cost per user was near zero, and can actually be offset by the network effect benefits. In the AI era however, marginal cost per user scales linearly, and there are little to no network effects. If all 3 billion Facebook users become Muse users, this would cost Meta about $2.5 trillion a year. Meta is therefore pressured to monetise on every user. This means that every user must become a “high stakes user”, ie use Muse for real purchases.

The business model in the AI era is completely different from that of Web era. Traditional internet companies thrive on user quantity, making it a competition for MAU’s. This was because the marginal cost per user was near zero, and can actually be offset by the network effect benefits. In the AI era however, marginal cost per user scales linearly, and there are little to no network effects. If all 3 billion Facebook users become Muse users, this would cost Meta about $2.5 trillion a year. Meta is therefore pressured to monetise on every user. This means that every user must become a “high stakes user”, ie use Muse for real purchases.

GeLun Ding

703,548 views • 1 month ago

Boris Cherny recorded this internally for his team. It was never meant to go public. He built Claude Code and shows how to go from idea to production-ready product in just 30 minutes. No script. No cuts. He just talks and types. This is the most practical video on Claude Code I’ve seen. Watch the video, then read about loops below.

Boris Cherny recorded this internally for his team. It was never meant to go public. He built Claude Code and shows how to go from idea to production-ready product in just 30 minutes. No script. No cuts. He just talks and types. This is the most practical video on Claude Code I’ve seen. Watch the video, then read about loops below.

Raytar

13,787 views • 4 months ago

5 patterns in Text Arena's price–performance Pareto frontier since 2023: 1. GPT-4-level quality is now ~500x lower cost. - From a ~$50 blended price per million tokens in 2023 to ~$0.10 today. 2. The higher-price end is both better and lower-priced since 2023. - The leading Arena score has climbed ~170 points (1,330 → 1,500). While the price of the higher-end frontier models dropped from ~$50 to ~$20 per million tokens. 3. The low-cost end gained the most. - Under $0.20 per million tokens, the best available model went from ~1,000 Arena score in 2023 to ~1,440 today. 4. The low-cost/top performance gap has nearly closed. - In 2023, sub-$0.20 models trailed the leader by ~350 Arena points. Today, ~60. 5. The cast has rotated quite a bit. - - OpenAI set the 2023–24 benchmark. - AI at Meta strengthened the low-cost end in 2024. - Google DeepMind drove the 2025 jump. - Anthropic holds the peak in 2026. - xAI and Chinese labs like DeepSeek AI, Z.ai, Kimi.ai, Xiaomi MiMo, and Qwen are continuing to push the mid-price frontier.

5 patterns in Text Arena's price–performance Pareto frontier since 2023: 1. GPT-4-level quality is now ~500x lower cost. - From a ~$50 blended price per million tokens in 2023 to ~$0.10 today. 2. The higher-price end is both better and lower-priced since 2023. - The leading Arena score has climbed ~170 points (1,330 → 1,500). While the price of the higher-end frontier models dropped from ~$50 to ~$20 per million tokens. 3. The low-cost end gained the most. - Under $0.20 per million tokens, the best available model went from ~1,000 Arena score in 2023 to ~1,440 today. 4. The low-cost/top performance gap has nearly closed. - In 2023, sub-$0.20 models trailed the leader by ~350 Arena points. Today, ~60. 5. The cast has rotated quite a bit. - - OpenAI set the 2023–24 benchmark. - AI at Meta strengthened the low-cost end in 2024. - Google DeepMind drove the 2025 jump. - Anthropic holds the peak in 2026. - xAI and Chinese labs like DeepSeek AI, Z.ai, Kimi.ai, Xiaomi MiMo, and Qwen are continuing to push the mid-price frontier.

Arena.ai

58,356 views • 5 months ago

MiniMax H3 Just Dropped on Lovart! 🔥 The world's #2 video model just landed on Lovart. Production-quality AI video at less than half the cost. •15s cinematic output with fluid, high-fidelity motion •15multimodal references: text, image, video & audio •Consistent characters, products, scenes & styles From Ads to MV to TVC, no more choosing between quality and cost.

MiniMax H3 Just Dropped on Lovart! 🔥 The world's #2 video model just landed on Lovart. Production-quality AI video at less than half the cost. •15s cinematic output with fluid, high-fidelity motion •15multimodal references: text, image, video & audio •Consistent characters, products, scenes & styles From Ads to MV to TVC, no more choosing between quality and cost.

LovartAI

241,649 views • 3 months ago

Google has just unveiled Veo 2, its latest cutting-edge AI video model. In performance tests, Veo outperformed OpenAI's Sora in both video quality and accuracy to prompts. The compilation below was generated entirely by AI - Full tutorial on our AI campus- link in bio #Ai

Google has just unveiled Veo 2, its latest cutting-edge AI video model. In performance tests, Veo outperformed OpenAI's Sora in both video quality and accuracy to prompts. The compilation below was generated entirely by AI - Full tutorial on our AI campus- link in bio #Ai

Daryl Tufekci

21,143 views • 1 year ago

GPT-5.2-high by OpenAI is off to a strong start in the Code Arena. ⚡️ If you’re new here: Code Arena is where AI models build full web apps, tools, and interactive sites — all from a single prompt. Watch the video to see GPT-5.2-high in action, then try your own prompt and reply with your creation below. ⬇️

GPT-5.2-high by OpenAI is off to a strong start in the Code Arena. ⚡️ If you’re new here: Code Arena is where AI models build full web apps, tools, and interactive sites — all from a single prompt. Watch the video to see GPT-5.2-high in action, then try your own prompt and reply with your creation below. ⬇️

Arena.ai

246,430 views • 11 months ago

Man In the Arena is the story about just your average man. One person's fight against the world. Tune in on at 3pmBST May 22nd for the full music video.

Man In the Arena is the story about just your average man. One person's fight against the world. Tune in on at 3pmBST May 22nd for the full music video.

the script

12,570 views • 5 months ago

The first ever CLAUDE MADE ANIME is out!! It's called Gus & Spark and 25 minute episode. Opus 5.5 MAX worked on it for over 15 hours, and wrote it code by code with 0 AI generated video. - 34K lines of code - 986M tokens, about $370 at API prices I believe Fable 5.5 should be able to produce even higher quality anime! In the future all 3D animated videos can be created by Claude without any AI generated video.

The first ever CLAUDE MADE ANIME is out!! It's called Gus & Spark and 25 minute episode. Opus 5.5 MAX worked on it for over 15 hours, and wrote it code by code with 0 AI generated video. - 34K lines of code - 986M tokens, about $370 at API prices I believe Fable 5.5 should be able to produce even higher quality anime! In the future all 3D animated videos can be created by Claude without any AI generated video.

Ziwen

365,991 views • 24 days ago

🚨ELON: CYBERCAB WILL MAKE AUTONOMOUS TRANSPORT AS CHEAP AS 20 CENTS PER MILE "I think the cost of autonomous transport will be so low, it’ll feel like individualized mass transit. The average cost of a bus for a city is about $1 per mile, but we expect the CyberCab’s operating cost to be around 20 cents per mile, and including taxes, it’ll be 30 or 40 cents per mile. And yes, you’ll be able to buy one for below $30,000." Source: Tesla Elon Musk

🚨ELON: CYBERCAB WILL MAKE AUTONOMOUS TRANSPORT AS CHEAP AS 20 CENTS PER MILE "I think the cost of autonomous transport will be so low, it’ll feel like individualized mass transit. The average cost of a bus for a city is about $1 per mile, but we expect the CyberCab’s operating cost to be around 20 cents per mile, and including taxes, it’ll be 30 or 40 cents per mile. And yes, you’ll be able to buy one for below $30,000." Source: Tesla Elon Musk

Mario Nawfal

721,119 views • 2 years ago

We just cut AI spend by 75% across an entire workforce in one click. Introducing: Merge for Workforce. Without Merge, your employees are left choosing one of two options: 1. Spend 100x the cost they should be from using frontier models for everything 2. Deliver subpar work from overly restrictive model access Instead with Merge, IT can connect any identity provider and set model routing policies by team in one click. Merge pushes them to every machine through a desktop client, overriding the model configs across your team’s AI assistants and coding tools. Now every task gets the right model for the job. The result? 75x fewer tokens with faster and better output.

We just cut AI spend by 75% across an entire workforce in one click. Introducing: Merge for Workforce. Without Merge, your employees are left choosing one of two options: 1. Spend 100x the cost they should be from using frontier models for everything 2. Deliver subpar work from overly restrictive model access Instead with Merge, IT can connect any identity provider and set model routing policies by team in one click. Merge pushes them to every machine through a desktop client, overriding the model configs across your team’s AI assistants and coding tools. Now every task gets the right model for the job. The result? 75x fewer tokens with faster and better output.

Shensi Ding

525,727 views • 2 months ago

We just launched AI Motion Graphics with Anthropic Think vibecoding for motion design. The cost of professional motion work just dropped to zero. All generated from a single prompt. Small teams can now produce the same quality as large agencies. No After Effects, no templates, no code — just describe what you want. Try it on

We just launched AI Motion Graphics with Anthropic Think vibecoding for motion design. The cost of professional motion work just dropped to zero. All generated from a single prompt. Small teams can now produce the same quality as large agencies. No After Effects, no templates, no code — just describe what you want. Try it on

Invideo

466,257 views • 9 months ago

AI at Meta is back in the game >Muse Wonder at xhigh reasoning , animation is purely code > this is super fast model , like it feels like i am using a fast mode voices by elevenlabs and music by suno decide by muse congrats Alexandr Wang Mark Zuckerberg

AI at Meta is back in the game >Muse Wonder at xhigh reasoning , animation is purely code > this is super fast model , like it feels like i am using a fast mode voices by elevenlabs and music by suno decide by muse congrats Alexandr Wang Mark Zuckerberg

Chetaslua

43,946 views • 28 days ago

muse wonder vs swe 2 tested both models with the same prompt at highest reasoning available but came out really different > muse took 24 mins and cost about $18 > swe took 15 mins and cost zero dollars really surprised by how good swe 2 did here but i don't know what muse wonder was doing model is scored above astra and fable 5.1 but it’s performing way worse than both in tests

muse wonder vs swe 2 tested both models with the same prompt at highest reasoning available but came out really different > muse took 24 mins and cost about $18 > swe took 15 mins and cost zero dollars really surprised by how good swe 2 did here but i don't know what muse wonder was doing model is scored above astra and fable 5.1 but it’s performing way worse than both in tests

J A Z I I

69,319 views • 28 days ago

we created a one-minute lore accurate video with a single slack message from the toilet. it's not to the quality of our bespoke work yet, but it's changing the way we think about AI content in lore-rich worlds. ⬇️ learn about how Showrunner™ works and see the "prompt" below

we created a one-minute lore accurate video with a single slack message from the toilet. it's not to the quality of our bespoke work yet, but it's changing the way we think about AI content in lore-rich worlds. ⬇️ learn about how Showrunner™ works and see the "prompt" below

NEUKO

12,390 views • 6 months ago

In the past 6 months... - Muse Spark - Muse Image - Muse Video - Muse Spark 1.1 - Meta Model API - Muse Glimmer - Muse Spark 1.2 - Muse Code - Muse Spark 1.3 - Muse ...and we’re just getting started. 🚀

In the past 6 months... - Muse Spark - Muse Image - Muse Video - Muse Spark 1.1 - Meta Model API - Muse Glimmer - Muse Spark 1.2 - Muse Code - Muse Spark 1.3 - Muse ...and we’re just getting started. 🚀

AI at Meta

192,468 views • 1 month ago

There isn’t one model that’s best at everything anymore. Choosing the right one for each task (without sacrificing quality or paying more than you need to) is a hard engineering problem. That’s what we’re working on with Auto in Copilot. Today, more than 90% of Copilot queries use Auto to dynamically route work across models and optimize for quality, speed, and cost. In this video, I share how it works and chooses the right model for the job...

There isn’t one model that’s best at everything anymore. Choosing the right one for each task (without sacrificing quality or paying more than you need to) is a hard engineering problem. That’s what we’re working on with Auto in Copilot. Today, more than 90% of Copilot queries use Auto to dynamically route work across models and optimize for quality, speed, and cost. In this video, I share how it works and chooses the right model for the job...

Charles Lamanna

71,471 views • 10 days ago

AI is using more and more of the world’s energy. But how much economic value are we getting from AI? We’re introducing Joules-to-GDP (J2G), a framework that measures how well AI turns energy into economic value and shows which tasks, models, and hardware give the best return. Across 44 occupations in finance, engineering, law, administrative work, and more in OpenAI's GDPval, we arrive at three findings: 1) The tasks with the highest economic value are not the tasks with the highest economic value per joule. For example, CAD design earns ~10% more per task than building websites, but the latter earns ~27% more value per joule. 2) The most powerful model isn’t always the best use of energy. Gemma 4 31B delivers 56% of GLM 5.2’s net value with 23% of the energy, for 2.5x the value per joule. 3) More chips don’t guarantee proportionate returns. Moving the same model from 1 to 8 NVIDIA H100s uses 3.2x the energy but cuts completion time by only 22%, with about the same value. With Snorkel AI, we’re also releasing GDPWorks, a benchmark for economically valuable tasks across 36 additional occupations. Joint work co-led by Amir Zeinali and Avanika Narayan, with Mersad, Jon Hartley, John Hennessy, Hazy Research: Strip Mall AI Research Club, and Azalia Mirhoseini. More details below 🧵

AI is using more and more of the world’s energy. But how much economic value are we getting from AI? We’re introducing Joules-to-GDP (J2G), a framework that measures how well AI turns energy into economic value and shows which tasks, models, and hardware give the best return. Across 44 occupations in finance, engineering, law, administrative work, and more in OpenAI's GDPval, we arrive at three findings: 1) The tasks with the highest economic value are not the tasks with the highest economic value per joule. For example, CAD design earns ~10% more per task than building websites, but the latter earns ~27% more value per joule. 2) The most powerful model isn’t always the best use of energy. Gemma 4 31B delivers 56% of GLM 5.2’s net value with 23% of the energy, for 2.5x the value per joule. 3) More chips don’t guarantee proportionate returns. Moving the same model from 1 to 8 NVIDIA H100s uses 3.2x the energy but cuts completion time by only 22%, with about the same value. With Snorkel AI, we’re also releasing GDPWorks, a benchmark for economically valuable tasks across 36 additional occupations. Joint work co-led by Amir Zeinali and Avanika Narayan, with Mersad, Jon Hartley, John Hennessy, Hazy Research: Strip Mall AI Research Club, and Azalia Mirhoseini. More details below 🧵

Jon Saad-Falcon

167,682 views • 5 days ago

meta muse just mogged fable 5.1 tested fable 5.1 and muse spark 1.3 with same prompt at highest reasoning available and results came out really different > muse spark 1.3 completed task in one minute and costed almost nothing > fable 5.1 completed task in 70 minutes and costed $13 it's clear that meta is stepping up their game in ai, they don't wanna left behind which one did better here?

meta muse just mogged fable 5.1 tested fable 5.1 and muse spark 1.3 with same prompt at highest reasoning available and results came out really different > muse spark 1.3 completed task in one minute and costed almost nothing > fable 5.1 completed task in 70 minutes and costed $13 it's clear that meta is stepping up their game in ai, they don't wanna left behind which one did better here?

J A Z I I

75,566 views • 2 months ago