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.

NVIDIA did it again.. they trained a GPT that generates human movement instead of text.. next-token prediction, but the tokens are body motions. it's called GPC and it hit a 99.98% success rate reproducing a massive corpus of motion clips.. cartwheels, vaults, flips, all of it. → 99.98% success... rate reproducing motion clips → learns cartwheels, flips, vaulting from raw data → adapts to new tasks with <1% extra parameters studios spend 6-12 months and millions of dollars building custom controllers for one character in one game. GPC does it in one forward pass. Pixar, Ubisoft, and every mocap studio on the planet just felt a chill down their spine. the same next-token trick that gave us ChatGPT is now doing parkour inside a physics engine.show more

How To Prompt

36,763 subscribers

102,590 views • 2 months ago •via X (Twitter)

Science & Technology

Anya Rossi• Live Now

Private livecam show

29 Comments

Chen Tessler2 months ago

Fully open-source --

How To Prompt2 months ago

Paper:

Albert Renshaw2 months ago

Nothing more than stochastic parkour

Husni Marwan2 months ago

I want to see them collaboratively plays football

Alex 3yb0^2 months ago

Great! How can I use that with my rigged characters in Blender ?

Frosty402 months ago

felt a chill down their spline* fixed it for you

🧩2 months ago

so it's just a basic movement diffusion model?

capt2 months ago

Every time people say transformers have reached their limit, someone applies the same idea to a completely different domain. I had a similar reaction after spending time with Claude Opus 5 - it wasn't one feature that impressed me, it was how much more generally capable the model felt. If this trend continues, motion generation is probably just another step, not the destination.

Da Bitcoin Bandit2 months ago

great now install the model in a kill bot

VND2 months ago

All this technology to make devs jobs EASIER and CHEAPER to produce things yet the prices are absolutely astronomical for games. Make it make sense.

theH2 months ago

cool!

未知2 months ago

说实话，99.98%的动作重现成功率确实亮眼，但这恰恰暴露了当前AI在物理世界理解上的致命短板：它只是在像素或骨骼点层面做“下一个标记预测”，本质上跟GPT生成文本没有区别，根本不理解重力、动量、肌肉发力这些物理约束。GPC能完美复现侧手翻，但换个没见过的动作组合可能就崩了。NVIDIA用Transformer暴力堆算力解决动作生成，思路很直接，但离真正的具身智能还差一个物理引擎的距离。这更像是高级动作数据库的插值检索，而不是智能体��…

Trevor West2 months ago

As interesting as this is, it was the wrong approach... No human moves perfectly, the real money is in imperfect movement (a perfect model is step one only). The same goes for speech in regards to communication tech. using AI.

ata2 months ago

love how committed nvidia is to the whole stack. i will love my future humanoid being able to do these neat tricks

Abdulmuiz Adeyemo2 months ago

Looks cool

Sebastian Buzdugan2 months ago

i've seen motion models pass replay then fail on foot contact in rollout

Mente Prompt • Inteligencia Artificial2 months ago

great paper from nvidia

CreatAlpha2 months ago

@grok 这个论文说的什么？能预测机器人的下一个动作？

expulvar2 months ago

why do i feel tired for the thing?

Alexander Petrov2 months ago

So, will a robot actually move around in real world, using it?

bitmin2 months ago

When NPC's created their own NPC's, what happened to the game?

byMAR.CO2 months ago

Give it a physical body

Anna A. Sobczak2 months ago

Does it adapt to different body geometries?

takeshi fujita2 months ago

、

Aryan2 months ago

this will be super useful for building high fidelity simulations

BlueSfaira1 month ago

This is so big for the gaming industry! Could this be mapped to humanoid robots?

Tommaso Bianco2 months ago

If this is the resulting video, does not seem completely natural to me. Besides movements, gravity

PlainSightLogic2 months ago

If it can be applied to this, it can be applied to almost anything.

Corsyphus 🚀2 months ago

Fifa this immediately on my Xbox!

Related Videos

MotionGPT: Human Motion as a Foreign Language paper page: Though the advancement of pre-trained large language models unfolds, the exploration of building a unified model for language and other multi-modal data, such as motion, remains challenging and untouched so far. Fortunately, human motion displays a semantic coupling akin to human language, often perceived as a form of body language. By fusing language data with large-scale motion models, motion-language pre-training that can enhance the performance of motion-related tasks becomes feasible. Driven by this insight, we propose MotionGPT, a unified, versatile, and user-friendly motion-language model to handle multiple motion-relevant tasks. Specifically, we employ the discrete vector quantization for human motion and transfer 3D motion into motion tokens, similar to the generation process of word tokens. Building upon this "motion vocabulary", we perform language modeling on both motion and text in a unified manner, treating human motion as a specific language. Moreover, inspired by prompt learning, we pre-train MotionGPT with a mixture of motion-language data and fine-tune it on prompt-based question-and-answer tasks. Extensive experiments demonstrate that MotionGPT achieves state-of-the-art performances on multiple motion tasks including text-driven motion generation, motion captioning, motion prediction, and motion in-between.

MotionGPT: Human Motion as a Foreign Language paper page: Though the advancement of pre-trained large language models unfolds, the exploration of building a unified model for language and other multi-modal data, such as motion, remains challenging and untouched so far. Fortunately, human motion displays a semantic coupling akin to human language, often perceived as a form of body language. By fusing language data with large-scale motion models, motion-language pre-training that can enhance the performance of motion-related tasks becomes feasible. Driven by this insight, we propose MotionGPT, a unified, versatile, and user-friendly motion-language model to handle multiple motion-relevant tasks. Specifically, we employ the discrete vector quantization for human motion and transfer 3D motion into motion tokens, similar to the generation process of word tokens. Building upon this "motion vocabulary", we perform language modeling on both motion and text in a unified manner, treating human motion as a specific language. Moreover, inspired by prompt learning, we pre-train MotionGPT with a mixture of motion-language data and fine-tune it on prompt-based question-and-answer tasks. Extensive experiments demonstrate that MotionGPT achieves state-of-the-art performances on multiple motion tasks including text-driven motion generation, motion captioning, motion prediction, and motion in-between.

AK

125,353 views • 3 years ago

Everyone is chasing that 100x token. Right now $tibbir is my pick. But it's not just a Hail Mary. It's a bet on one of the most successful firms in the industry. I break down just some of the compelling evidence suggesting Ribbit Capital is behind the $tibbir token and what it could mean if they are.

Everyone is chasing that 100x token. Right now $tibbir is my pick. But it's not just a Hail Mary. It's a bet on one of the most successful firms in the industry. I break down just some of the compelling evidence suggesting Ribbit Capital is behind the $tibbir token and what it could mean if they are.

Bulldog1205

33,346 views • 1 year ago

Ansem says $VVV is one of the sleeper consumer app play in crypto "VVV is a sleeper consumer app play onchain. I think VVV is one of the sleeper plays in crypto" "They have like 4 million users right now. I think they just hit like $100 million annualized revenue" "VVV is one of the applications that has found breakout success and consumer success, which is extremely hard to do with customers that aren't even really crypto-native" "I think the narrative for them as uncensored AI is just great generally going forward" "For the same reason that Bitcoin made a ton of sense, like uncensored money, you should have uncensored access to AI, especially if AI is going to continue to get a lot more impressive over the years" "It's a great trade. It was a really strong leader this year. It went from $1 to $20" "Their numbers are fucking through the roof. They're crushing it" "I think it's rare that a crypto app is crushing it outside of crypto and also has a token attached to it. And the team is focused on delivering value to the token as well" "I like Venice a lot. There's not really much that they're competing with in the AI space right now in crypto"

Ansem says $VVV is one of the sleeper consumer app play in crypto "VVV is a sleeper consumer app play onchain. I think VVV is one of the sleeper plays in crypto" "They have like 4 million users right now. I think they just hit like $100 million annualized revenue" "VVV is one of the applications that has found breakout success and consumer success, which is extremely hard to do with customers that aren't even really crypto-native" "I think the narrative for them as uncensored AI is just great generally going forward" "For the same reason that Bitcoin made a ton of sense, like uncensored money, you should have uncensored access to AI, especially if AI is going to continue to get a lot more impressive over the years" "It's a great trade. It was a really strong leader this year. It went from $1 to $20" "Their numbers are fucking through the roof. They're crushing it" "I think it's rare that a crypto app is crushing it outside of crypto and also has a token attached to it. And the team is focused on delivering value to the token as well" "I like Venice a lot. There's not really much that they're competing with in the AI space right now in crypto"

Shadooow

17,144 views • 1 month ago

Two data points dropped in the last few months that should terrify every software company that thinks its codebase is a moat. First, one engineer at Cloudflare, working with Claude via AI agents, rebuilt 94% of Next.js, one of the most widely used frontend frameworks on the internet, built over 10 years by a large engineering team in a single week. Total cost was $1,100 in API tokens. The result, called Vinext, is a drop-in replacement that builds production apps up to 4x faster and produces client bundles 57% smaller and customers are already running it in production. Second is Cursor CEO Michael Truell deployed a swarm of hundreds of GPT-5.2 agents that ran uninterrupted for an entire week and built a fully functional web browser from scratch called FastRender. 3 million lines of code, thousands of files and a custom Rust rendering engine with HTML parsing, CSS layout, text shaping, and a custom JavaScript VM. Total cost was roughly $30,000. For context, Google has spent billions of dollars and decades of engineering building Chrome. And the benchmarks say by next year, you will be able to one-shot prompt anything. The moat that software companies spent decades building, the complexity of their codebase, the years it would take a competitor to replicate it, the switching costs that moat assumed humans were the unit of production. AI does not care how long it took you to build it, it only cares how long it takes to rebuild it. And right now, the answer is one week.

Two data points dropped in the last few months that should terrify every software company that thinks its codebase is a moat. First, one engineer at Cloudflare, working with Claude via AI agents, rebuilt 94% of Next.js, one of the most widely used frontend frameworks on the internet, built over 10 years by a large engineering team in a single week. Total cost was $1,100 in API tokens. The result, called Vinext, is a drop-in replacement that builds production apps up to 4x faster and produces client bundles 57% smaller and customers are already running it in production. Second is Cursor CEO Michael Truell deployed a swarm of hundreds of GPT-5.2 agents that ran uninterrupted for an entire week and built a fully functional web browser from scratch called FastRender. 3 million lines of code, thousands of files and a custom Rust rendering engine with HTML parsing, CSS layout, text shaping, and a custom JavaScript VM. Total cost was roughly $30,000. For context, Google has spent billions of dollars and decades of engineering building Chrome. And the benchmarks say by next year, you will be able to one-shot prompt anything. The moat that software companies spent decades building, the complexity of their codebase, the years it would take a competitor to replicate it, the switching costs that moat assumed humans were the unit of production. AI does not care how long it took you to build it, it only cares how long it takes to rebuild it. And right now, the answer is one week.

Milk Road AI

16,781 views • 5 months ago

I don't think people understand that one of the biggest real-world use cases for Crypto it is happening - right now. FedEx, just sat down with Mance Harmon of $HBAR and explained why they chose Hedera to track millions of packages a day, to 220 countries. Dale Christie, Blockchain Strategist at FedEx said in 2018, FedEx tried building its own blockchain... But it wouldn’t scale so they chose Hedera. Why? Every package generates hundreds - sometimes thousands - of data points, and Hedera enables the key facts to move through customs at the speed of data, not paper. “When you are a trusted shipper and I am a trusted receiver… that data can move through customs at the speed of data and keep it moving.” One of the largest logistics companies in the world just publicly said Hedera is the foundation they’re building on. Hedera is the most undervalued token imo. and this is one of the biggest use cases Crypto's ever seen.

I don't think people understand that one of the biggest real-world use cases for Crypto it is happening - right now. FedEx, just sat down with Mance Harmon of $HBAR and explained why they chose Hedera to track millions of packages a day, to 220 countries. Dale Christie, Blockchain Strategist at FedEx said in 2018, FedEx tried building its own blockchain... But it wouldn’t scale so they chose Hedera. Why? Every package generates hundreds - sometimes thousands - of data points, and Hedera enables the key facts to move through customs at the speed of data, not paper. “When you are a trusted shipper and I am a trusted receiver… that data can move through customs at the speed of data and keep it moving.” One of the largest logistics companies in the world just publicly said Hedera is the foundation they’re building on. Hedera is the most undervalued token imo. and this is one of the biggest use cases Crypto's ever seen.

Mark

44,717 views • 6 months ago

MIT PhD student Alex Zhang reveals the scaling result where a model trained on short tasks generalizes to problems 100x longer for free: "If you're very clever about the design of your harness or how you use the language model, you can almost get scaling gains for free." "If you train a model naively, there's no tricks. It's just the same way you train a model on these RL environments. You just roll it out, and then you just get some reward." "If you train it on only short tasks, like only tasks that are 10,000 tokens long, and then you were to run it on a similar domain, but at a million tokens, or 10 million tokens, or 100,000 tokens, it generalizes really, really well. If you look at it compared to even the base transformer, you get way better generalization properties." "When the model uses an RLM (Recursive Language Model) after it's trained on these short tasks, it will see some kind of trajectory of actions that it does. Between these two problems of different lengths, the RLM learns to see them as almost the same problem." "Token for token, they're almost the same. You can describe it in code. In one code setting, maybe the for loop is a little bigger, but it's the same kind of code and it derives the constants from the data. There's no hard coding, so they literally look the same." alex zhang

MIT PhD student Alex Zhang reveals the scaling result where a model trained on short tasks generalizes to problems 100x longer for free: "If you're very clever about the design of your harness or how you use the language model, you can almost get scaling gains for free." "If you train a model naively, there's no tricks. It's just the same way you train a model on these RL environments. You just roll it out, and then you just get some reward." "If you train it on only short tasks, like only tasks that are 10,000 tokens long, and then you were to run it on a similar domain, but at a million tokens, or 10 million tokens, or 100,000 tokens, it generalizes really, really well. If you look at it compared to even the base transformer, you get way better generalization properties." "When the model uses an RLM (Recursive Language Model) after it's trained on these short tasks, it will see some kind of trajectory of actions that it does. Between these two problems of different lengths, the RLM learns to see them as almost the same problem." "Token for token, they're almost the same. You can describe it in code. In one code setting, maybe the for loop is a little bigger, but it's the same kind of code and it derives the constants from the data. There's no hard coding, so they literally look the same." alex zhang

MTS

100,519 views • 2 months ago

“Next-token prediction” just got a serious rival 🤯 Ant Group just dropped LLaDA 2.1, and it challenges the dominant paradigm of LLMs. Unlike most models that generate one token at a time, LLaDA 2.1 uses diffusion to generate blocks of text in parallel. Why this changes everything: → Global Planning: It effectively sees the "future" while writing the "past" → Parallel Generation: It generates chunks in parallel, not sequentially, being much faster → Massive Efficiency: 16B MoE architecture that only uses ~1.4B active parameters per step. 100% Open Source.

“Next-token prediction” just got a serious rival 🤯 Ant Group just dropped LLaDA 2.1, and it challenges the dominant paradigm of LLMs. Unlike most models that generate one token at a time, LLaDA 2.1 uses diffusion to generate blocks of text in parallel. Why this changes everything: → Global Planning: It effectively sees the "future" while writing the "past" → Parallel Generation: It generates chunks in parallel, not sequentially, being much faster → Massive Efficiency: 16B MoE architecture that only uses ~1.4B active parameters per step. 100% Open Source.

Alvaro Cintas

31,277 views • 7 months ago

Notch Clips. every other ad tool ignores the footage sitting in your dropbox and generates new clips from a prompt. notch opens your library instead. it watches every clip you have ever shot and understands what is inside each one. the mood. who is in frame. what they are doing. the quality. so when you ask it to build an ad, it does not start from scratch. it reaches into your past shoots, finds the clip that fits, and slots it into the exact moment that needs it. here is what that looks like. you had thirteen versions of one reaction shot. notch picked the one that worked and left the other twelve. not by filename. not by guessing from a reference image. it watched all thirteen and could tell the difference. that is the whole idea. the right clip was already in your library. every other tool was about to make a worse one. footage you already shot, turned into every ad you still need. drop your site and a few clips you have. notch builds the first ad from them.

Notch Clips. every other ad tool ignores the footage sitting in your dropbox and generates new clips from a prompt. notch opens your library instead. it watches every clip you have ever shot and understands what is inside each one. the mood. who is in frame. what they are doing. the quality. so when you ask it to build an ad, it does not start from scratch. it reaches into your past shoots, finds the clip that fits, and slots it into the exact moment that needs it. here is what that looks like. you had thirteen versions of one reaction shot. notch picked the one that worked and left the other twelve. not by filename. not by guessing from a reference image. it watched all thirteen and could tell the difference. that is the whole idea. the right clip was already in your library. every other tool was about to make a worse one. footage you already shot, turned into every ad you still need. drop your site and a few clips you have. notch builds the first ad from them.

Vinay Jain

406,286 views • 3 months ago

Not a preplanned motion sequence. A robot deciding mid-jump what to do next. [📍 paper + demo] Researchers just showed a humanoid doing real parkour using only onboard perception. No motion script, no fixed obstacle layout. The system is called Perceptive Humanoid Parkour (PHP). Instead of memorizing a path, the robot reads depth from its cameras and continuously chooses actions. Step, vault, climb, or roll depending on what geometry appears in front of it. To make that possible, they combine three ideas: First, they stitch together human motion clips into long movement references so the robot learns fluid transitions instead of isolated tricks. Second, they train tracking policies with reinforcement learning so contacts land at the right time and the robot keeps balance during dynamic moves. Finally, everything is distilled into one perception policy that runs directly from depth input to action selection. The result on a Unitree G1: about 3 m/s vaults wall climbs up to 1.25 m nearly one minute continuous obstacle traversal adapting when obstacles move What matters is not the tricks. It is the shift in capability. Earlier humanoids executed motions. This one navigates situations. Once robots react to geometry instead of replaying trajectories, environments stop needing to be predictable. Warehouses, homes, and outdoors suddenly become the same problem. Thanks for sharing, Zhen Wu! Paper + demo: ——— Weekly robotics and AI insights. Subscribe free:

Not a preplanned motion sequence. A robot deciding mid-jump what to do next. [📍 paper + demo] Researchers just showed a humanoid doing real parkour using only onboard perception. No motion script, no fixed obstacle layout. The system is called Perceptive Humanoid Parkour (PHP). Instead of memorizing a path, the robot reads depth from its cameras and continuously chooses actions. Step, vault, climb, or roll depending on what geometry appears in front of it. To make that possible, they combine three ideas: First, they stitch together human motion clips into long movement references so the robot learns fluid transitions instead of isolated tricks. Second, they train tracking policies with reinforcement learning so contacts land at the right time and the robot keeps balance during dynamic moves. Finally, everything is distilled into one perception policy that runs directly from depth input to action selection. The result on a Unitree G1: about 3 m/s vaults wall climbs up to 1.25 m nearly one minute continuous obstacle traversal adapting when obstacles move What matters is not the tricks. It is the shift in capability. Earlier humanoids executed motions. This one navigates situations. Once robots react to geometry instead of replaying trajectories, environments stop needing to be predictable. Warehouses, homes, and outdoors suddenly become the same problem. Thanks for sharing, Zhen Wu! Paper + demo: ——— Weekly robotics and AI insights. Subscribe free:

Ilir Aliu

22,127 views • 7 months ago

Satya Nadella, CEO of Microsoft, says every company now runs on two kinds of capital, and leaking one of them is a one-way door: "This economy is going to be shaped going forward by both human capital and, let's call it, this token capital. That is true for Microsoft, that is true for a new startup, that's going to be true for any bank that's been in existence for 100 years." "The interplay between human capital and token capital, and compounding the returns of it, is what you need to do." "What is the tacit knowledge of an enterprise or a firm? It's the unique ways that you're able to operate, pass judgment, have taste." "The model, in some sense, is able to extract that tacit knowledge through human trajectories, and then code it in a set of weights in a model." "Every enterprise now needs to be more mindful about that interplay of humans and their digital estate working together, those trajectories training the models that they keep as IP versus leak it." "Because if you leak it, it's a one-way door. You're done in some sense." The moat was never the data sitting in your warehouse. It was the judgment your people apply to it, and AI now turns that judgment into weights you either own or hand away. Treat your tacit knowledge like a patent, not a prompt. The firms that pipe a hundred years of taste into someone else's frontier model are trading a permanent edge for a quarter of convenience.

Satya Nadella, CEO of Microsoft, says every company now runs on two kinds of capital, and leaking one of them is a one-way door: "This economy is going to be shaped going forward by both human capital and, let's call it, this token capital. That is true for Microsoft, that is true for a new startup, that's going to be true for any bank that's been in existence for 100 years." "The interplay between human capital and token capital, and compounding the returns of it, is what you need to do." "What is the tacit knowledge of an enterprise or a firm? It's the unique ways that you're able to operate, pass judgment, have taste." "The model, in some sense, is able to extract that tacit knowledge through human trajectories, and then code it in a set of weights in a model." "Every enterprise now needs to be more mindful about that interplay of humans and their digital estate working together, those trajectories training the models that they keep as IP versus leak it." "Because if you leak it, it's a one-way door. You're done in some sense." The moat was never the data sitting in your warehouse. It was the judgment your people apply to it, and AI now turns that judgment into weights you either own or hand away. Treat your tacit knowledge like a patent, not a prompt. The firms that pipe a hundred years of taste into someone else's frontier model are trading a permanent edge for a quarter of convenience.

Karl Mehta

34,573 views • 2 months ago

Someone on Reddit just built a AAA-quality snow simulation in their browser. Using Claude Opus 5. Millions of snow particles. A character in a robe with cloth physics. Five water-based spells that carve the terrain. A snow surfing system with a wake that throws spray into sunlight. Every footstep displaces snow. Trails form berms at the edges. Trenches slowly refill over time. Spells leave permanent craters. The entire thing runs in one browser tab. WebGPU. No game engine. No 3D modeling software. Just code. He wrote one detailed prompt. Opus 5 planned the architecture, wrote the shaders, built the physics, iterated from screenshots, and documented every decision. 9 hours. ~4 million tokens. You can open it right now and walk around in it. This isn't a demo. This is the kind of thing a small studio would spend months building.

Someone on Reddit just built a AAA-quality snow simulation in their browser. Using Claude Opus 5. Millions of snow particles. A character in a robe with cloth physics. Five water-based spells that carve the terrain. A snow surfing system with a wake that throws spray into sunlight. Every footstep displaces snow. Trails form berms at the edges. Trenches slowly refill over time. Spells leave permanent craters. The entire thing runs in one browser tab. WebGPU. No game engine. No 3D modeling software. Just code. He wrote one detailed prompt. Opus 5 planned the architecture, wrote the shaders, built the physics, iterated from screenshots, and documented every decision. 9 hours. ~4 million tokens. You can open it right now and walk around in it. This isn't a demo. This is the kind of thing a small studio would spend months building.

Vaibhav Sisinty

1,219,935 views • 1 month ago

3Blue1Brown’s new video explains why every LLM is actually a compression machine. everyone describes pre-training as “next token prediction” but that’s just the surface-level objective. in reality it is a means to making the most efficient text compressor. prediction and compression are two sides of the same coin. when you train the model to predict the next token you’re not just teaching it to guess the next word but how to best encode the human knowledge it sees. better compression means better abstraction means better reasoning at some point, compression stops looking like storage or a database (as some like to call it on X) and looks like an approximation of understanding.

3Blue1Brown’s new video explains why every LLM is actually a compression machine. everyone describes pre-training as “next token prediction” but that’s just the surface-level objective. in reality it is a means to making the most efficient text compressor. prediction and compression are two sides of the same coin. when you train the model to predict the next token you’re not just teaching it to guess the next word but how to best encode the human knowledge it sees. better compression means better abstraction means better reasoning at some point, compression stops looking like storage or a database (as some like to call it on X) and looks like an approximation of understanding.

ℏεsam

120,145 views • 3 months ago

OpenAI. said. this. publicly. their own engineers just proved one idea on themselves, in writing: stop telling AI what's wrong. hand it the whole broken thing and let it find out they gave GPT-6 Astra a slow test build of their own coding tool. one cause found, a memory bottleneck, one allocator swapped, every turn 25× faster this is GPT-6 Astra, the layer that fixes the cause instead of the symptom, $0 on top of the ChatGPT plan you already pay for: - open ChatGPT or Codex, pick GPT-6 Astra, hand it the whole thing: the folder, the file that takes a minute to open. it works in apps with no API and reads your screen - type one sentence: find the one cause, prove it, fix it, do not patch around it - leave the room. it asks without stopping, keeps working on what does not need your answer, waits only where the answer changes the outcome - keep it in one Codex session with the experimental notes setting on: it remembers across context windows why an earlier fix failed - expect the first pass to land: handed a program with no source, it worked out how it runs 88% of the time first try, 99.2% within four you never find out what was broken. it gets fixed anyway the catch is on the same page. roughly 30% more memory for that speed, and the safety checks can pause a long job until you approve the next step describing the problem was the expensive half of fixing it. that half just ended every hour you spend explaining the symptom to a chat window, someone else has handed theirs over whole bookmark this before the next thing breaks, the playbook for handing a whole job to an AI worker is in the piece below ↓

OpenAI. said. this. publicly. their own engineers just proved one idea on themselves, in writing: stop telling AI what's wrong. hand it the whole broken thing and let it find out they gave GPT-6 Astra a slow test build of their own coding tool. one cause found, a memory bottleneck, one allocator swapped, every turn 25× faster this is GPT-6 Astra, the layer that fixes the cause instead of the symptom, $0 on top of the ChatGPT plan you already pay for: - open ChatGPT or Codex, pick GPT-6 Astra, hand it the whole thing: the folder, the file that takes a minute to open. it works in apps with no API and reads your screen - type one sentence: find the one cause, prove it, fix it, do not patch around it - leave the room. it asks without stopping, keeps working on what does not need your answer, waits only where the answer changes the outcome - keep it in one Codex session with the experimental notes setting on: it remembers across context windows why an earlier fix failed - expect the first pass to land: handed a program with no source, it worked out how it runs 88% of the time first try, 99.2% within four you never find out what was broken. it gets fixed anyway the catch is on the same page. roughly 30% more memory for that speed, and the safety checks can pause a long job until you approve the next step describing the problem was the expensive half of fixing it. that half just ended every hour you spend explaining the symptom to a chat window, someone else has handed theirs over whole bookmark this before the next thing breaks, the playbook for handing a whole job to an AI worker is in the piece below ↓

Argona

107,134 views • 14 days ago

Rune Christensen, Co-Founder of Sky (Sky), on USDS, STUSDS, and risk tiers in DeFi "Lindy, like stablecoin product with a really really reasonable rate of return" "right now it's at 3.75%, which is actually 10 points above SOFR" "a lot of people dunking on DeFi for not being able to beat TradFi, but USDS does in fact still beat TradFi and is very much highly risk optimized" "with Sky you can also use a product like STUSDS, which is a higher risk token that still has this link to the US dollar" "you can actually go even more high risk and buy something like SPK, which is a completely volatile governance token of one of the Sky agents" "STUSDS, because it's a risk token and it's explicitly marketed as risk capital, it's for advanced users because it has higher risk" "it's not just a stablecoin that is designed to protect you above everything else" "with STUSDS it's also about being able to offer a higher rate of return" "one of the consequences of this is that it can have its liquidity, its one to one liquidity with USDS frozen"

Rune Christensen, Co-Founder of Sky (Sky), on USDS, STUSDS, and risk tiers in DeFi "Lindy, like stablecoin product with a really really reasonable rate of return" "right now it's at 3.75%, which is actually 10 points above SOFR" "a lot of people dunking on DeFi for not being able to beat TradFi, but USDS does in fact still beat TradFi and is very much highly risk optimized" "with Sky you can also use a product like STUSDS, which is a higher risk token that still has this link to the US dollar" "you can actually go even more high risk and buy something like SPK, which is a completely volatile governance token of one of the Sky agents" "STUSDS, because it's a risk token and it's explicitly marketed as risk capital, it's for advanced users because it has higher risk" "it's not just a stablecoin that is designed to protect you above everything else" "with STUSDS it's also about being able to offer a higher rate of return" "one of the consequences of this is that it can have its liquidity, its one to one liquidity with USDS frozen"

Frontliner 💯

75,006 views • 4 months ago

is genspark ($50M ARR in 6mo) underrated next to chatgpt, claude, gemini, and manus? i spent 34 minutes testing its “super ai agent” and how it actually works. 1/ you write one prompt and it runs through gpt-5, claude sonnet, gemini 2.5, and mistral large then gives you the best response. 2/ you upload one image and it runs it through nano banana, flux, and bytedance then returns the most realistic version. 3/ you generate a video and genspark routes it through wan 2.2, pika labs, gen-2, and runway ml combining the best clips and motion. 4/ you paste an idea for slides and it uses gpt-5, claude 3 opus, gemini advanced, and genspark’s slide model to create a full deck with visuals and market data. 5/ AI slides and building a fundraising deck with a few prompts (wasn't perfect, but it was a good start for my personal workflow) what's cool about genspark is it's inexpensive and does it all. kinda like the costco of LLMs. TLDR; you hand it one prompt + input, and genspark routes sub-tasks across models and tools behind the scenes. it’s orchestration over one ideal model. perspective shift. instead of jumping from model to model, llm to llm, genspark shows what the next generation of ai will look like orchestration over obsession. and for the rest of us using it, the only thing that matters is getting the best outcome. what other AI app should I review next? i'll keep doing it if it's helpful for folks.

is genspark ($50M ARR in 6mo) underrated next to chatgpt, claude, gemini, and manus? i spent 34 minutes testing its “super ai agent” and how it actually works. 1/ you write one prompt and it runs through gpt-5, claude sonnet, gemini 2.5, and mistral large then gives you the best response. 2/ you upload one image and it runs it through nano banana, flux, and bytedance then returns the most realistic version. 3/ you generate a video and genspark routes it through wan 2.2, pika labs, gen-2, and runway ml combining the best clips and motion. 4/ you paste an idea for slides and it uses gpt-5, claude 3 opus, gemini advanced, and genspark’s slide model to create a full deck with visuals and market data. 5/ AI slides and building a fundraising deck with a few prompts (wasn't perfect, but it was a good start for my personal workflow) what's cool about genspark is it's inexpensive and does it all. kinda like the costco of LLMs. TLDR; you hand it one prompt + input, and genspark routes sub-tasks across models and tools behind the scenes. it’s orchestration over one ideal model. perspective shift. instead of jumping from model to model, llm to llm, genspark shows what the next generation of ai will look like orchestration over obsession. and for the rest of us using it, the only thing that matters is getting the best outcome. what other AI app should I review next? i'll keep doing it if it's helpful for folks.

GREG ISENBERG

62,172 views • 11 months ago

$Introducing Immutable Audience: the growth engine for pre-launch Steam games. (btw we didn’t spend $50k on a viral launch agency - so if you think this is cool just share it for free i guess) Game marketing is broken. Every year PC studios spend billions to grow their games, and yet: - they don’t know which channels are actually working - only a fraction of wishlist sources can be tracked - of those people who DO wishlist games, the rate of those who go on to purchase is falling After 8 years building games ourselves, and working with over 700 studios, the need for something new was obvious. So, we built Audience. How it works: 1. Audience pulls in real time data from your players across Steam, social media, ad platforms, influencers and more 2. It stores this marketing data in your player CRM, and uses our Enhanced Attribution to identify the one channel driving 80% of your wishlists 3. Finally, it generates a custom growth audit, reviewing your real time data against winning strategies from hundreds of games to identify the most effective ways to get you wishlists each week During Audience’s beta we’ve already seen incredible results. We helped one game grow by over 300k wishlists in 2 months, and then more than *doubled* the rate of those players who converted into paying customers, compared to relying on Steam alone. For the last few months the team at Immutable has been heads down building. We’re incredibly grateful to our community for your support as we onboard the world. Slowly, then all at once. To celebrate the launch, we’re offering one month of Audience for free to any studio that signs up by commenting on this video. And for the first 100 studios that comment, we’ll create you a custom growth plan for free. Comment “Audience” and we’ll send it through.$

Introducing Immutable Audience: the growth engine for pre-launch Steam games. (btw we didn’t spend $50k on a viral launch agency - so if you think this is cool just share it for free i guess) Game marketing is broken. Every year PC studios spend billions to grow their games, and yet: - they don’t know which channels are actually working - only a fraction of wishlist sources can be tracked - of those people who DO wishlist games, the rate of those who go on to purchase is falling After 8 years building games ourselves, and working with over 700 studios, the need for something new was obvious. So, we built Audience. How it works: 1. Audience pulls in real time data from your players across Steam, social media, ad platforms, influencers and more 2. It stores this marketing data in your player CRM, and uses our Enhanced Attribution to identify the one channel driving 80% of your wishlists 3. Finally, it generates a custom growth audit, reviewing your real time data against winning strategies from hundreds of games to identify the most effective ways to get you wishlists each week During Audience’s beta we’ve already seen incredible results. We helped one game grow by over 300k wishlists in 2 months, and then more than doubled the rate of those players who converted into paying customers, compared to relying on Steam alone. For the last few months the team at Immutable has been heads down building. We’re incredibly grateful to our community for your support as we onboard the world. Slowly, then all at once. To celebrate the launch, we’re offering one month of Audience for free to any studio that signs up by commenting on this video. And for the first 100 studios that comment, we’ll create you a custom growth plan for free. Comment “Audience” and we’ll send it through.

Robbie Ferguson | Immutable

24,819 views • 1 month ago

This work makes a humanoid robot do simple parkour moves by looking with a depth camera and choosing the right move on the fly. The big deal is that it turns lots of small human moves into long, real-time robot behavior, without hand-coding every transition or retraining for each new course. A humanoid robot is usually good at steady walking, but it often fails when it has to do fast moves like jumping up, vaulting, or rolling, and then keep going to the next obstacle. The hard part is that you cannot easily collect training data for every possible obstacle shape, distance, and mistake, so robots end up learning a few moves that only work in a narrow setup. This work starts from short clips of real human parkour moves, like stepping over, vaulting, climbing, and rolling. It uses motion matching, which is basically a smart “pick the next clip that fits best right now” search, to stitch those short clips into a long, smooth plan that looks like a human doing a whole course. Then it trains a controller with reinforcement learning (RL), which means the robot learns by trial and error to copy that plan while staying balanced and not falling. After training separate expert controllers for different moves, it compresses them into 1 controller that uses only onboard depth sensing and a simple “go this fast in this direction” command. In real tests on a Unitree G1 humanoid, it can clear multiple obstacles in a row, adapt when obstacles get moved, and climb a wall up to 1.25m.

This work makes a humanoid robot do simple parkour moves by looking with a depth camera and choosing the right move on the fly. The big deal is that it turns lots of small human moves into long, real-time robot behavior, without hand-coding every transition or retraining for each new course. A humanoid robot is usually good at steady walking, but it often fails when it has to do fast moves like jumping up, vaulting, or rolling, and then keep going to the next obstacle. The hard part is that you cannot easily collect training data for every possible obstacle shape, distance, and mistake, so robots end up learning a few moves that only work in a narrow setup. This work starts from short clips of real human parkour moves, like stepping over, vaulting, climbing, and rolling. It uses motion matching, which is basically a smart “pick the next clip that fits best right now” search, to stitch those short clips into a long, smooth plan that looks like a human doing a whole course. Then it trains a controller with reinforcement learning (RL), which means the robot learns by trial and error to copy that plan while staying balanced and not falling. After training separate expert controllers for different moves, it compresses them into 1 controller that uses only onboard depth sensing and a simple “go this fast in this direction” command. In real tests on a Unitree G1 humanoid, it can clear multiple obstacles in a row, adapt when obstacles get moved, and climb a wall up to 1.25m.

Rohan Paul

37,121 views • 7 months ago

Why the character movement in my custom game engine felt janky and how I fixed it. In a game engine, most often, a character moves using the physics engine. Meaning, the player is not just a coordinate in space but a physical body. It has velocity, it handles collisions, and it interacts with the world. Now, as you might know, physics engines need stability. If you run them at variable framerates, things start breaking. Objects phase through walls or fly off into space because the math becomes unpredictable. This is why most game engines lock their physics loop to a 60Hz fixed rate. But here’s the problem: If you have a high-end system, you don't want to limit it at 60 FPS. That's a waste of good hardware. Now, that said, if the GPU is rendering at 144 FPS but the player's position (physics driven) only updates 60 times a second, it creates a micro-stutter that ruins the "smooth" feel of the game. A good way to fix this is to treat the character as two separate things: 1. The Physics Body (Invisible part): This is the "real" character. It lives in the 60Hz physics world, it moves the player and handles collisions. 2. The Visual Model and Camera (Visible part): This is what the player actually sees. It doesn't care about collisions, its only job is to look nice and smooth at whatever framerate the GPU is pushing. Once you have this separation, you can use interpolation to keep them in sync. Every time the physics clock ticks, you save the previous position of the invisible body before moving it to the new one. Between those ticks, calculate how far we are between the last physics update and the next one. By using this to drive the visible parts of the game, the stutters disappear. The physics loop stays fixed behind the scenes, while the visuals slide smoothly between the snapshots. Example: - Right after a tick: blend_weight= 0.0 (The visual model stays at the old physics position). - Halfway to the next: blend_weight= 0.5 (The visual model slides to the middle point). - Just before the next: blend_weight= 0.9 (The visual model is almost at the new physics position). Pro-Tip A critical mistake I made initially, and one many devs make, is parenting the camera and visible parts directly to the player body. If you do this, the camera inherits the discrete 60Hz physics movement by default. In that setup, interpolation won't work because the camera is "stuck" to the physics clock. For this fix to work you must decouple the camera and visuals from the body and move them separately. Player movement processing in Detis Engine: - fixed_process: Physics runs at 60Hz. Handles collisions and raw movement. - process: Variable rate. Mainly used for player input caching in the player case. - late_process: Variable rate. Handles interpolated camera movement after physics and everything else is done being processed. - render. Submits the final interpolated transforms to the GPU. The test environment in the video is running on an old 2070-based laptop. Hopefully the video compression won't introduce any stutter... I’m sharing this in hopes it helps a fellow dev. Cheers.

Why the character movement in my custom game engine felt janky and how I fixed it. In a game engine, most often, a character moves using the physics engine. Meaning, the player is not just a coordinate in space but a physical body. It has velocity, it handles collisions, and it interacts with the world. Now, as you might know, physics engines need stability. If you run them at variable framerates, things start breaking. Objects phase through walls or fly off into space because the math becomes unpredictable. This is why most game engines lock their physics loop to a 60Hz fixed rate. But here’s the problem: If you have a high-end system, you don't want to limit it at 60 FPS. That's a waste of good hardware. Now, that said, if the GPU is rendering at 144 FPS but the player's position (physics driven) only updates 60 times a second, it creates a micro-stutter that ruins the "smooth" feel of the game. A good way to fix this is to treat the character as two separate things: 1. The Physics Body (Invisible part): This is the "real" character. It lives in the 60Hz physics world, it moves the player and handles collisions. 2. The Visual Model and Camera (Visible part): This is what the player actually sees. It doesn't care about collisions, its only job is to look nice and smooth at whatever framerate the GPU is pushing. Once you have this separation, you can use interpolation to keep them in sync. Every time the physics clock ticks, you save the previous position of the invisible body before moving it to the new one. Between those ticks, calculate how far we are between the last physics update and the next one. By using this to drive the visible parts of the game, the stutters disappear. The physics loop stays fixed behind the scenes, while the visuals slide smoothly between the snapshots. Example: - Right after a tick: blend_weight= 0.0 (The visual model stays at the old physics position). - Halfway to the next: blend_weight= 0.5 (The visual model slides to the middle point). - Just before the next: blend_weight= 0.9 (The visual model is almost at the new physics position). Pro-Tip A critical mistake I made initially, and one many devs make, is parenting the camera and visible parts directly to the player body. If you do this, the camera inherits the discrete 60Hz physics movement by default. In that setup, interpolation won't work because the camera is "stuck" to the physics clock. For this fix to work you must decouple the camera and visuals from the body and move them separately. Player movement processing in Detis Engine: - fixed_process: Physics runs at 60Hz. Handles collisions and raw movement. - process: Variable rate. Mainly used for player input caching in the player case. - late_process: Variable rate. Handles interpolated camera movement after physics and everything else is done being processed. - render. Submits the final interpolated transforms to the GPU. The test environment in the video is running on an old 2070-based laptop. Hopefully the video compression won't introduce any stutter... I’m sharing this in hopes it helps a fellow dev. Cheers.

Ioannis Koukourakis

48,703 views • 8 months ago

NEW OPEN SOURCE ROBOT: A cheap, lightweight, and agile platform for locomotion! Called BRIDGE, it is an open-source humanoid robot platform, hardware plus control system. It is an 88 cm, 12.5 kg humanoid with 21 active degrees of freedom that costs roughly $1,500. Its distinguishing idea is that the body and the controller are designed together through a three-stage data-driven pipeline: - anatomical constraints - kinematic criteria - closed-loop dynamic criteria Hence the morphology itself is optimized to preserve human motion quality when the robot executes. It demonstrates teleoperated locomotion, balance-oriented motion (bending, lunging, extreme balance) and highly dynamic motion including back-flips and expressive sequences. I like that the thesis is designing the body and the controller together, not a controller for a fixed body, which inverts how nearly everyone else works; the robot is usually given, usually a Unitree G1, then train a policy on it. At the end of the day, it is a sub-$2K open-source humanoid that does back-flips, the humanoid analogue of the cheap-arm democratization.

NEW OPEN SOURCE ROBOT: A cheap, lightweight, and agile platform for locomotion! Called BRIDGE, it is an open-source humanoid robot platform, hardware plus control system. It is an 88 cm, 12.5 kg humanoid with 21 active degrees of freedom that costs roughly $1,500. Its distinguishing idea is that the body and the controller are designed together through a three-stage data-driven pipeline: - anatomical constraints - kinematic criteria - closed-loop dynamic criteria Hence the morphology itself is optimized to preserve human motion quality when the robot executes. It demonstrates teleoperated locomotion, balance-oriented motion (bending, lunging, extreme balance) and highly dynamic motion including back-flips and expressive sequences. I like that the thesis is designing the body and the controller together, not a controller for a fixed body, which inverts how nearly everyone else works; the robot is usually given, usually a Unitree G1, then train a policy on it. At the end of the day, it is a sub-$2K open-source humanoid that does back-flips, the humanoid analogue of the cheap-arm democratization.

Léo

23,248 views • 5 days ago