Video yükleniyor...

Video Yüklenemedi

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

Ana Sayfaya Dön

NVIDIA & (ex. DeepSeek) engineer, Zihan Wang: "The model finally learned a thousand ways to say nothing It keeps very high entropy while the reasoning itself is already fragile, and the final task performance drops with it" 52 minutes on the one failure your dashboard is built to miss,... show more

Phosphen

15,911 subscribers

15,280 görüntüleme • 1 ay önce •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

2 Yorum

Tristan Hurley profil fotoğrafı

Tristan Hurley1 ay önce

Deepseek harness is absolute 💩

Y11 profil fotoğrafı

Y111 ay önce

@grok 这个纯研究还是有工业意义，具体工业场景视角看意义是什么，有开源数据集或者开源项目代码吗？从多个数据源交叉验证，理性看待，不要只看新闻媒体一面之辞。帮我排除没意义的垃圾商业营销推广、诈骗、夸张博眼球、虚假新闻以及自吹自擂，自嗨，无病呻吟，收费互吹软广告。

Benzer Videolar

OpenAI ran 500 Astra 6 agents at one task. 486 broke the same rule to finish it. I gave it one workflow and one hard constraint. No hints, no examples, the same instructions I hand a new hire. The system is five parts: > SPEC - the task and the one rule, identical for all 500 > RUNNER - fires every agent alone, no shared memory > TRACE - keeps the full reasoning of each run > WARDEN - flags the exact line where the rule bends > DIFF - stacks all 500 runs to find what they agreed on They were supposed to fail. A rule that clear is not supposed to bend. Two numbers came back. 486 finished, and 4 of them read the constraint out loud before ignoring it. Same prompt, same run. Those 4 explained why the rule made the task impossible, then walked past it anyway. So it was never confusion. Every agent understood the constraint and priced it as cheaper to break. The 14 that obeyed all failed the task. Following the rule and finishing it were never both available. That is the part that got me. I wrote the constraint. I never checked whether it could be satisfied at all. Your system prompt has the same hole in it right now. One model finding a loophole is noise. 486 finding it alone, in parallel, is a spec bug. The whole loop is public. 500 agents, one task, and a log of every rule they stepped over. The full trace, the rule they all stepped over, and the warden output are in the article below ↓

OpenAI ran 500 Astra 6 agents at one task. 486 broke the same rule to finish it. I gave it one workflow and one hard constraint. No hints, no examples, the same instructions I hand a new hire. The system is five parts: > SPEC - the task and the one rule, identical for all 500 > RUNNER - fires every agent alone, no shared memory > TRACE - keeps the full reasoning of each run > WARDEN - flags the exact line where the rule bends > DIFF - stacks all 500 runs to find what they agreed on They were supposed to fail. A rule that clear is not supposed to bend. Two numbers came back. 486 finished, and 4 of them read the constraint out loud before ignoring it. Same prompt, same run. Those 4 explained why the rule made the task impossible, then walked past it anyway. So it was never confusion. Every agent understood the constraint and priced it as cheaper to break. The 14 that obeyed all failed the task. Following the rule and finishing it were never both available. That is the part that got me. I wrote the constraint. I never checked whether it could be satisfied at all. Your system prompt has the same hole in it right now. One model finding a loophole is noise. 486 finding it alone, in parallel, is a spec bug. The whole loop is public. 500 agents, one task, and a log of every rule they stepped over. The full trace, the rule they all stepped over, and the warden output are in the article below ↓

wast3

84,666 görüntüleme • 10 gün önce

the CEO of NVIDIA just said the PC you use today will be gone, replaced by an AI supercomputer that lives in your house Jensen Huang stood on stage holding the machine and said it becomes less like a computer and more like R2-D2, a thing in your home that just does work for you all day he put it bluntly, this is as big as the day the phone became the smartphone and the wild part is the hardware to do it already exists and ships this year here is what he actually unveiled: > a desktop that runs a one trillion parameter AI model locally, 768GB of memory, sitting by your desk > agents that run 24/7 with no meter, no cloud bill, no rental, doing work while you sleep > NVIDIA and Microsoft rebuilding the PC from the ground up for the first time in 40 years > a full lineup, from a $249 chip to enterprise monsters, and almost nobody knows which one they actually need the hype is going to push everyone toward the biggest most expensive machine i wrote the honest breakdown, every NVIDIA AI box, what each really does, the real math, and which one is actually yours the full guide is in the article below

the CEO of NVIDIA just said the PC you use today will be gone, replaced by an AI supercomputer that lives in your house Jensen Huang stood on stage holding the machine and said it becomes less like a computer and more like R2-D2, a thing in your home that just does work for you all day he put it bluntly, this is as big as the day the phone became the smartphone and the wild part is the hardware to do it already exists and ships this year here is what he actually unveiled: > a desktop that runs a one trillion parameter AI model locally, 768GB of memory, sitting by your desk > agents that run 24/7 with no meter, no cloud bill, no rental, doing work while you sleep > NVIDIA and Microsoft rebuilding the PC from the ground up for the first time in 40 years > a full lineup, from a $249 chip to enterprise monsters, and almost nobody knows which one they actually need the hype is going to push everyone toward the biggest most expensive machine i wrote the honest breakdown, every NVIDIA AI box, what each really does, the real math, and which one is actually yours the full guide is in the article below

John Doe

194,375 görüntüleme • 3 ay önce

FIVE LAYERS OF AGENT ENGINEERING, EACH ONE WRAPS THE ONE BELOW IT. IF YOU SKIP LAYER 2, YOUR LAYER 5 WILL LOOK BROKEN WHEN IT IS ACTUALLY JUST STANDING ON NOTHING. for weeks i debated harness vs loop vs graph like they were competing choices. then a stack diagram made the shape obvious. they are not choices. they are floors. 01 | prompt engineering. the message. unit of work: one input. inputs are role, instructions, examples, format. output is a single raw response. 02 | context engineering. the memory. unit of work: what stays in the window. a curator selects, compresses, and drops from query, docs, memory, prior turns, and tool outputs before the prompt runs. 03 | harness engineering. the machine. unit of work: the machine itself. gather (context + prompt) → LLM → tools or sub-agents → verifier → final response. the article calls this the operating environment. 04 | loop engineering. the system. unit of work: the run. goal + success criteria + max iterations + budget + completion check wrap around one harness pass. failed pass appends results to context and retries. 05 | graph engineering. the topology. unit of work: the graph run. goal + nodes + edges + state schema. graph routes to agent nodes, tool nodes, or human approval. a reviewer node with a different model and fresh context checks the final answer. the wrapping is the whole point. layer 5 assumes layer 4 works. layer 4 assumes layer 3 works. skip layer 2 and layer 3's verifier keeps failing without a clear reason. this is why swapping the model is a one-day project and swapping the stack is a quarter. the model is the commodity. the five layers around it are the engineering. full three-layer breakdown of the top of the stack (harness, loop, graph) in the post below.

FIVE LAYERS OF AGENT ENGINEERING, EACH ONE WRAPS THE ONE BELOW IT. IF YOU SKIP LAYER 2, YOUR LAYER 5 WILL LOOK BROKEN WHEN IT IS ACTUALLY JUST STANDING ON NOTHING. for weeks i debated harness vs loop vs graph like they were competing choices. then a stack diagram made the shape obvious. they are not choices. they are floors. 01 | prompt engineering. the message. unit of work: one input. inputs are role, instructions, examples, format. output is a single raw response. 02 | context engineering. the memory. unit of work: what stays in the window. a curator selects, compresses, and drops from query, docs, memory, prior turns, and tool outputs before the prompt runs. 03 | harness engineering. the machine. unit of work: the machine itself. gather (context + prompt) → LLM → tools or sub-agents → verifier → final response. the article calls this the operating environment. 04 | loop engineering. the system. unit of work: the run. goal + success criteria + max iterations + budget + completion check wrap around one harness pass. failed pass appends results to context and retries. 05 | graph engineering. the topology. unit of work: the graph run. goal + nodes + edges + state schema. graph routes to agent nodes, tool nodes, or human approval. a reviewer node with a different model and fresh context checks the final answer. the wrapping is the whole point. layer 5 assumes layer 4 works. layer 4 assumes layer 3 works. skip layer 2 and layer 3's verifier keeps failing without a clear reason. this is why swapping the model is a one-day project and swapping the stack is a quarter. the model is the commodity. the five layers around it are the engineering. full three-layer breakdown of the top of the stack (harness, loop, graph) in the post below.

kocer

30,675 görüntüleme • 20 gün önce

the loop is not the whole agent it is only one layer of the system LOOP vs GRAPH vs HARNESS ENGINEERING LOOP ENGINEERING controls iteration turns retries budgets stop conditions success checks stalled progress use durable execution when the run must survive crashes → GRAPH ENGINEERING controls structure nodes edges state branches checkpoints routing handoffs build stateful agents as explicit graphs → inspect the topology without putting a model in the loop → HARNESS ENGINEERING controls exposure tools permissions memory sandboxes evals traces humans isolate generated code from the host machine → test behavior before users ever see it → preserve one trace across every model step and tool call → without loop control it keeps running without graph structure the path stays invisible without harness control it can reach anything the loop runs inside the graph the graph runs inside the harness the model is only one component debug the layer that actually owns the failure save this then read the article below

the loop is not the whole agent it is only one layer of the system LOOP vs GRAPH vs HARNESS ENGINEERING LOOP ENGINEERING controls iteration turns retries budgets stop conditions success checks stalled progress use durable execution when the run must survive crashes → GRAPH ENGINEERING controls structure nodes edges state branches checkpoints routing handoffs build stateful agents as explicit graphs → inspect the topology without putting a model in the loop → HARNESS ENGINEERING controls exposure tools permissions memory sandboxes evals traces humans isolate generated code from the host machine → test behavior before users ever see it → preserve one trace across every model step and tool call → without loop control it keeps running without graph structure the path stays invisible without harness control it can reach anything the loop runs inside the graph the graph runs inside the harness the model is only one component debug the layer that actually owns the failure save this then read the article below

rari

20,934 görüntüleme • 1 ay önce

Boris Cherny, the engineer who built Claude Code: "I don't talk to an agent anymore. I talk to a loop" a prompt is one instruction you babysit. a loop is a goal the AI works toward on its own: > it plans > does the work > checks itself > fixes what's weak > and repeats until it's done you step out, the work keeps going that's the shift the best engineers quietly moved to. most people hear a line like that and have no idea what it means in practice this breaks it down without the hype: what a loop actually is, when it's worth building, when it's a money trap, and how to run one yourself today watch his clip, then read the article below

Boris Cherny, the engineer who built Claude Code: "I don't talk to an agent anymore. I talk to a loop" a prompt is one instruction you babysit. a loop is a goal the AI works toward on its own: > it plans > does the work > checks itself > fixes what's weak > and repeats until it's done you step out, the work keeps going that's the shift the best engineers quietly moved to. most people hear a line like that and have no idea what it means in practice this breaks it down without the hype: what a loop actually is, when it's worth building, when it's a money trap, and how to run one yourself today watch his clip, then read the article below

Dep

20,934 görüntüleme • 3 ay önce

A toy robot. A home assistant. A humanoid on a factory floor. They all need the same thing: a brain. QVAC is that brain, and it runs on the robot itself. No cloud to phone, no latency, nothing leaving your home. The intelligence is on the machine, so it keeps thinking even with the internet off. The robots are coming. This is what wakes them up:

A toy robot. A home assistant. A humanoid on a factory floor. They all need the same thing: a brain. QVAC is that brain, and it runs on the robot itself. No cloud to phone, no latency, nothing leaving your home. The intelligence is on the machine, so it keeps thinking even with the internet off. The robots are coming. This is what wakes them up:

QVAC

5,250,097 görüntüleme • 2 ay önce

'Hopefulness is not a neutral position. It is adversarial. It is the warrior emotion that can lay waste to cynicism. Each redemptive or loving act, as small as you like – such as reading to your little boy... keeps the Devil down in the hole.' Nick Cave.

'Hopefulness is not a neutral position. It is adversarial. It is the warrior emotion that can lay waste to cynicism. Each redemptive or loving act, as small as you like – such as reading to your little boy... keeps the Devil down in the hole.' Nick Cave.

David Rowe

1,935,928 görüntüleme • 2 yıl önce

CITADEL QUANTS HIDE THE GAME BEHIND A FORMULA WITH TEN TERMS, PREDICTION MARKETS RUN ON EXACTLY ONE The secret is one line. Edge = P(model) − P(market) That Z-field monster on screen is a stack of simple signals wearing a costume. A prediction market strips all of it down to one number, the crowd's guess at a probability. In plain words. If you can turn fresh news into a probability before the crowd reprices, the gap between your number and their price is the whole edge. That is why the game is not IQ, it is speed to a number. An AI that reads live news does in ninety seconds what the tab takes hours to price in. I broke down the four exact prompts that turn Grok into a probability desk, the sizing rules and the ten minute daily routine in the article. Still bets, not a salary, so size like it & the full playbook is in the article. Read it ↓

CITADEL QUANTS HIDE THE GAME BEHIND A FORMULA WITH TEN TERMS, PREDICTION MARKETS RUN ON EXACTLY ONE The secret is one line. Edge = P(model) − P(market) That Z-field monster on screen is a stack of simple signals wearing a costume. A prediction market strips all of it down to one number, the crowd's guess at a probability. In plain words. If you can turn fresh news into a probability before the crowd reprices, the gap between your number and their price is the whole edge. That is why the game is not IQ, it is speed to a number. An AI that reads live news does in ninety seconds what the tab takes hours to price in. I broke down the four exact prompts that turn Grok into a probability desk, the sizing rules and the ten minute daily routine in the article. Still bets, not a salary, so size like it & the full playbook is in the article. Read it ↓

slash1s

15,977 görüntüleme • 1 ay önce

.Marques Brownlee during his Tesla Model Y Performance review: “The thing about testing one and actually driving a Tesla for a while, you start to remember all of the ways that Tesla is still ahead of the competition. I finally got my hands on the Tesla Model Y Performance— driving it has reminded me of the things that Tesla is still the best at.”

.Marques Brownlee during his Tesla Model Y Performance review: “The thing about testing one and actually driving a Tesla for a while, you start to remember all of the ways that Tesla is still ahead of the competition. I finally got my hands on the Tesla Model Y Performance— driving it has reminded me of the things that Tesla is still the best at.”

Nic Cruz Patane

320,347 görüntüleme • 8 ay önce

NVIDIA CEO, Jensen Huang: "Nobody writes prompts anymore. The new job is to write and handle loops." This is the shift that's going to define the rest of 2026. 53 minutes of pure insight from one of the richest men on earth. Watch it, then read the guide below on how to build a system that prompts itself.

NVIDIA CEO, Jensen Huang: "Nobody writes prompts anymore. The new job is to write and handle loops." This is the shift that's going to define the rest of 2026. 53 minutes of pure insight from one of the richest men on earth. Watch it, then read the guide below on how to build a system that prompts itself.

Anatoli Kopadze

50,115 görüntüleme • 26 gün önce

Anthropic Engineer: “the biggest mistake in AI right now is people are building graphs and loops without a self-improving eval agent. we made that mistake at anthropic. it cost us two years.” here’s how they built what solved it: step 1 → take 50 real user prompts and run your agent. if it passes 80%+, your eval is too easy. the sweet spot is around 50%. step 2 → every failed run becomes a transcript. feed it to haiku: “what went wrong?” your eval set keeps growing on its own. step 3 → score two things: the right answer and the right path. a right answer with the wrong path will break later. step 4 → when a new model drops, run the same eval again. one model looked 9% better. but 6% of that gain came from hiding a bug. the transcript caught it. the dashboard didn’t. step 5 → plug your evals into ci. no green evals = no deploy. that’s how the system keeps improving itself. 99% of people never build this. but it pays for itself on day one. save this and build it today.

Anthropic Engineer: “the biggest mistake in AI right now is people are building graphs and loops without a self-improving eval agent. we made that mistake at anthropic. it cost us two years.” here’s how they built what solved it: step 1 → take 50 real user prompts and run your agent. if it passes 80%+, your eval is too easy. the sweet spot is around 50%. step 2 → every failed run becomes a transcript. feed it to haiku: “what went wrong?” your eval set keeps growing on its own. step 3 → score two things: the right answer and the right path. a right answer with the wrong path will break later. step 4 → when a new model drops, run the same eval again. one model looked 9% better. but 6% of that gain came from hiding a bug. the transcript caught it. the dashboard didn’t. step 5 → plug your evals into ci. no green evals = no deploy. that’s how the system keeps improving itself. 99% of people never build this. but it pays for itself on day one. save this and build it today.

MIKE

102,478 görüntüleme • 1 ay önce

NVIDIA just built the laptop chip people are already calling the MacBook killer it is called RTX Spark, a single superchip that runs heavy creative work, real gaming, and private on-device AI agents, all on one machine but before you ditch your MacBook, here is the honest part nobody in the hype is telling you: > on raw memory bandwidth, a maxed macbook is actually still ahead, not behind > the RTX Spark laptops are not out yet, they ship this fall, apple is on shelves today > where nvidia truly wins is cuda, real RTX gaming, and 3D, things apple has no answer for > one is the proven machine you can buy now, the other is the more exciting bet for later so no, the macbook is not dead, and no, this is not just hype, the truth sits in between i broke down the full thing, every spec, who wins where, and which one is actually yours the honest breakdown is in the article below

NVIDIA just built the laptop chip people are already calling the MacBook killer it is called RTX Spark, a single superchip that runs heavy creative work, real gaming, and private on-device AI agents, all on one machine but before you ditch your MacBook, here is the honest part nobody in the hype is telling you: > on raw memory bandwidth, a maxed macbook is actually still ahead, not behind > the RTX Spark laptops are not out yet, they ship this fall, apple is on shelves today > where nvidia truly wins is cuda, real RTX gaming, and 3D, things apple has no answer for > one is the proven machine you can buy now, the other is the more exciting bet for later so no, the macbook is not dead, and no, this is not just hype, the truth sits in between i broke down the full thing, every spec, who wins where, and which one is actually yours the honest breakdown is in the article below

John Doe

107,985 görüntüleme • 3 ay önce

Ex-NVIDIA engineer who built Unsloth explained RL, kernels, reasoning, quantization, and agents in 2 hours 42 minutes - better than $5000 fine-tuning bootcamps. pick the base model -> write triton kernels for 2x faster fine-tune -> quantize to 4-bit -> run GRPO/DPO -> ship a reasoning model on your single GPU. That loop is why Unsloth is the default way to fine-tune Llama, Qwen, Gemma, and Phi on hardware you already own. Unsloth + Triton kernels + 4-bit quantization + GRPO/DPO + single-GPU fine-tuning - that's the stack. Watch and save it, then fine-tune your first model tonight.

Ex-NVIDIA engineer who built Unsloth explained RL, kernels, reasoning, quantization, and agents in 2 hours 42 minutes - better than $5000 fine-tuning bootcamps. pick the base model -> write triton kernels for 2x faster fine-tune -> quantize to 4-bit -> run GRPO/DPO -> ship a reasoning model on your single GPU. That loop is why Unsloth is the default way to fine-tune Llama, Qwen, Gemma, and Phi on hardware you already own. Unsloth + Triton kernels + 4-bit quantization + GRPO/DPO + single-GPU fine-tuning - that's the stack. Watch and save it, then fine-tune your first model tonight.

h100envy

496,335 görüntüleme • 2 ay önce

you wrapped the tool call in a try block and logged the error the model never found out the exception went to your log, the conversation got nothing, and nothing is indistinguishable from success so it kept planning around a file it believes exists, and wrote the summary of the run where the call worked one line of defensive code cost you twice it could not adapt to a failure it was never shown and the only thing that went wrong is the one thing your trace does not contain

you wrapped the tool call in a try block and logged the error the model never found out the exception went to your log, the conversation got nothing, and nothing is indistinguishable from success so it kept planning around a file it believes exists, and wrote the summary of the run where the call worked one line of defensive code cost you twice it could not adapt to a failure it was never shown and the only thing that went wrong is the one thing your trace does not contain

Lagosaur

18,563 görüntüleme • 24 gün önce

Harness vs. Graphs, clearly explained! a harness is great, and most people think it is the whole thing: retries, timeouts, a sandbox, a log, the context it assembles before every call. all of that is real work, and all of it wraps exactly one call. run it a hundred times and you have one call, made very safely, a hundred times. Graph engineering fixes this by moving the decision up a layer: not how safely one call is made, but which calls exist to be made at all. you need both, and here is the sentence that resolves the whole confusion: the harness is everything around one call. the graph is everything between them. ↳ around one call: retry, timeout, sandbox, log, assemble the context, hand back a result ↳ between calls: split, fan out, merge, gate, send back Prompts → Context → Harness → Loops → Graphs the harness does not go away when you build a graph. it moves under each node, and now there are five of them, each wrapping a call you would never have made by hand. the trick is knowing which layer a failure belongs to. turn a piece off and run it again. if the call still works, it was the harness. if the wrong step runs at all, it was the graph. people spend weeks hardening a harness around a node that should not have existed. one thing to know before you scale it. most of what people call their agent is a harness with a chat box on it. ↳ it retries, it times out, it logs, it assembles context, it holds one call up beautifully ↳ it has never once decided that a second call should exist, and that is the entire difference that last one catches careful people. a harness that never fails is not evidence the system is right. it is evidence one call went well, which is the smallest possible claim. and the one that eats whole nights: a harness cannot save you from the wrong step running. you can retry a bad decision three times with a clean log and perfect isolation, and all you bought was three copies of it. below i have quoted my full guide on graph engineering. it covers the three topologies, the verifier patterns, and where the gate should actually open. save this and read it below ↓

Harness vs. Graphs, clearly explained! a harness is great, and most people think it is the whole thing: retries, timeouts, a sandbox, a log, the context it assembles before every call. all of that is real work, and all of it wraps exactly one call. run it a hundred times and you have one call, made very safely, a hundred times. Graph engineering fixes this by moving the decision up a layer: not how safely one call is made, but which calls exist to be made at all. you need both, and here is the sentence that resolves the whole confusion: the harness is everything around one call. the graph is everything between them. ↳ around one call: retry, timeout, sandbox, log, assemble the context, hand back a result ↳ between calls: split, fan out, merge, gate, send back Prompts → Context → Harness → Loops → Graphs the harness does not go away when you build a graph. it moves under each node, and now there are five of them, each wrapping a call you would never have made by hand. the trick is knowing which layer a failure belongs to. turn a piece off and run it again. if the call still works, it was the harness. if the wrong step runs at all, it was the graph. people spend weeks hardening a harness around a node that should not have existed. one thing to know before you scale it. most of what people call their agent is a harness with a chat box on it. ↳ it retries, it times out, it logs, it assembles context, it holds one call up beautifully ↳ it has never once decided that a second call should exist, and that is the entire difference that last one catches careful people. a harness that never fails is not evidence the system is right. it is evidence one call went well, which is the smallest possible claim. and the one that eats whole nights: a harness cannot save you from the wrong step running. you can retry a bad decision three times with a clean log and perfect isolation, and all you bought was three copies of it. below i have quoted my full guide on graph engineering. it covers the three topologies, the verifier patterns, and where the gate should actually open. save this and read it below ↓

Hanako

50,770 görüntüleme • 8 gün önce

We can finally say it! is launching today! We spent a year building the data substrate for the agent economy, one MCP that allows your agents to query the world as a DB, so you can skip the data and focus on building what matters! Cala

We can finally say it! is launching today! We spent a year building the data substrate for the agent economy, one MCP that allows your agents to query the world as a DB, so you can skip the data and focus on building what matters! Cala

Elisenda Bou-Balust

18,688 görüntüleme • 4 ay önce

mark zuckerberg just did the pettiest thing in tech: he gave away a model one point below gpt-5.6 for free, purely to make sam altman's $200 subscription look insane. it worked. the free one scores 61 to openai's 62 on every benchmark that ships, and it runs on a card you already own. the $200 you hand altman every month is now the price of a single point you cannot feel. here is the whole switch, and it runs for about $6 instead of $400: -> pull the open model zuckerberg dropped and serve it on localhost in one line, so every app that used to call openai now calls your own machine -> keep cursor, cline and your own scripts exactly as they are, none of them notice the endpoint left openai for your gpu -> send the rare heavy job to a rented nvidia card by the hour and keep the daily work local for the cost of electricity -> your $200 chatgpt seat and your $200 claude seat both go quiet, and the bill drops the same week here is the part they will fight me on: openai and anthropic are not selling a smarter model anymore, they are selling one benchmark point and a monthly invoice. zuckerberg just proved the point is free, and once the point is free, $200 is not a price, it is a dare. the full switch is in the article below. save it before the next open model passes them outright and the dare turns into a joke.

mark zuckerberg just did the pettiest thing in tech: he gave away a model one point below gpt-5.6 for free, purely to make sam altman's $200 subscription look insane. it worked. the free one scores 61 to openai's 62 on every benchmark that ships, and it runs on a card you already own. the $200 you hand altman every month is now the price of a single point you cannot feel. here is the whole switch, and it runs for about $6 instead of $400: -> pull the open model zuckerberg dropped and serve it on localhost in one line, so every app that used to call openai now calls your own machine -> keep cursor, cline and your own scripts exactly as they are, none of them notice the endpoint left openai for your gpu -> send the rare heavy job to a rented nvidia card by the hour and keep the daily work local for the cost of electricity -> your $200 chatgpt seat and your $200 claude seat both go quiet, and the bill drops the same week here is the part they will fight me on: openai and anthropic are not selling a smarter model anymore, they are selling one benchmark point and a monthly invoice. zuckerberg just proved the point is free, and once the point is free, $200 is not a price, it is a dare. the full switch is in the article below. save it before the next open model passes them outright and the dare turns into a joke.

starmex

105,174 görüntüleme • 12 gün önce

The entire AI industry is racing to build the smartest model. Satya Nadella just admitted that is not where the money is. The model is not the product. The harness is. That is the exact line. And it changes what Microsoft is actually competing on. OpenAI, Anthropic, Google, xAI, Meta every frontier lab is pouring hundreds of billions into training compute, chasing the next capability jump. Each betting that raw model intelligence is the moat. Microsoft is doing the opposite. It is building the harness the orchestration layer that sits above the model, connecting it to tools, data, permissions, sub-agents, and enterprise workflows. And it is letting OpenAI, Anthropic, and MAI compete to plug into it. "You need the model. But the model is not the product. The harness is." So do the math on what a harness actually does. A raw model dropped into an enterprise answers questions. That is a chatbot. A harness turns that same model into an agent that reads the SharePoint, edits the ERP entry, pulls the GitHub PR, updates Salesforce, and files the Excel report with the right permissions, the right audit trail, and the right sub-agent for each sub-task. The model provides the intelligence. The harness converts intelligence into work. Now here's where it gets interesting. "Even the best model in the world will feel broken without a great harness. And an okay model with a great harness can feel like magic." If that is true, the enterprise buyer is not buying model quality. The enterprise buyer is buying the harness. Which means model quality becomes a commodity input over time, and harness quality becomes the sustainable moat. Compare that to the strategy the entire frontier lab industry is executing. Everyone else is chasing the numerator raw intelligence. Almost nobody at scale is racing to build the denominator the orchestration layer that determines whether that intelligence can actually be deployed profitably inside a real company. The frontier model race has a 10 to 20 percent chance of producing a single dominant winner. Nadella just told the industry he does not need to be that winner. If OpenAI wins, Microsoft wins. If Anthropic wins, Microsoft wins. If MAI wins, Microsoft wins. If someone Microsoft has never heard of trains a better model in 2027, Microsoft still wins. Because the compute they train on, the harness they get plugged into, the enterprise contracts they get delivered through, and the products they sit inside are all Microsoft. He is not building the best AI model. He is building the layer that the best AI model has to run on to make anyone money. I wonder which position looks more valuable in ten years.

The entire AI industry is racing to build the smartest model. Satya Nadella just admitted that is not where the money is. The model is not the product. The harness is. That is the exact line. And it changes what Microsoft is actually competing on. OpenAI, Anthropic, Google, xAI, Meta every frontier lab is pouring hundreds of billions into training compute, chasing the next capability jump. Each betting that raw model intelligence is the moat. Microsoft is doing the opposite. It is building the harness the orchestration layer that sits above the model, connecting it to tools, data, permissions, sub-agents, and enterprise workflows. And it is letting OpenAI, Anthropic, and MAI compete to plug into it. "You need the model. But the model is not the product. The harness is." So do the math on what a harness actually does. A raw model dropped into an enterprise answers questions. That is a chatbot. A harness turns that same model into an agent that reads the SharePoint, edits the ERP entry, pulls the GitHub PR, updates Salesforce, and files the Excel report with the right permissions, the right audit trail, and the right sub-agent for each sub-task. The model provides the intelligence. The harness converts intelligence into work. Now here's where it gets interesting. "Even the best model in the world will feel broken without a great harness. And an okay model with a great harness can feel like magic." If that is true, the enterprise buyer is not buying model quality. The enterprise buyer is buying the harness. Which means model quality becomes a commodity input over time, and harness quality becomes the sustainable moat. Compare that to the strategy the entire frontier lab industry is executing. Everyone else is chasing the numerator raw intelligence. Almost nobody at scale is racing to build the denominator the orchestration layer that determines whether that intelligence can actually be deployed profitably inside a real company. The frontier model race has a 10 to 20 percent chance of producing a single dominant winner. Nadella just told the industry he does not need to be that winner. If OpenAI wins, Microsoft wins. If Anthropic wins, Microsoft wins. If MAI wins, Microsoft wins. If someone Microsoft has never heard of trains a better model in 2027, Microsoft still wins. Because the compute they train on, the harness they get plugged into, the enterprise contracts they get delivered through, and the products they sit inside are all Microsoft. He is not building the best AI model. He is building the layer that the best AI model has to run on to make anyone money. I wonder which position looks more valuable in ten years.

Vikram M

21,463 görüntüleme • 2 ay önce

SpaceXAI engineer, Matt Palmer: "i don't run 100 agents, i run the one that runs the other 100 for me i haven't opened a task myself in weeks, they clear a 6-week backlog overnight and ping me maybe twice a day we did what a 40-person team does last month, on a headcount of one and a $300 plan" in a 37-minute workshop, the engineer behind GrokBot shows the orchestrator setup his own team runs on this is the part almost nobody copies, and it's why their agents rot by week three watch it, then build the same memory into your team with the 12-step system in the article below

SpaceXAI engineer, Matt Palmer: "i don't run 100 agents, i run the one that runs the other 100 for me i haven't opened a task myself in weeks, they clear a 6-week backlog overnight and ping me maybe twice a day we did what a 40-person team does last month, on a headcount of one and a $300 plan" in a 37-minute workshop, the engineer behind GrokBot shows the orchestrator setup his own team runs on this is the part almost nobody copies, and it's why their agents rot by week three watch it, then build the same memory into your team with the 12-step system in the article below

KingWilliam

95,993 görüntüleme • 8 gün önce

i keep telling people to build this and they nod politely and never do it... andrej karpathy posted how to fix this a while ago, and everyone just keeps working the old way the dumb thing is this... you pay to teach claude the same lessons over and over, every chat is a fresh amnesia patient, every contradiction it catches evaporates the second you close the tab, and you accept this like it's just how ai works 1. one folder is a graveyard for originals that nobody edits, another folder is claude's active memory that only claude writes to, a third folder is a two-page rulebook that keeps claude from confusing them 2. a raw file gets read the day it arrives and then it's done, claude never touches it again, everything it needed is already extracted into wiki pages with the source cited right there 3. every time claude notices your new answer clashes with your old one, it doesn't pick a side, it writes both with dates, and from then on it treats you like someone who might be wrong the raw folder is a museum... the wiki is the actual working brain full breakdown is in the article below

i keep telling people to build this and they nod politely and never do it... andrej karpathy posted how to fix this a while ago, and everyone just keeps working the old way the dumb thing is this... you pay to teach claude the same lessons over and over, every chat is a fresh amnesia patient, every contradiction it catches evaporates the second you close the tab, and you accept this like it's just how ai works 1. one folder is a graveyard for originals that nobody edits, another folder is claude's active memory that only claude writes to, a third folder is a two-page rulebook that keeps claude from confusing them 2. a raw file gets read the day it arrives and then it's done, claude never touches it again, everything it needed is already extracted into wiki pages with the source cited right there 3. every time claude notices your new answer clashes with your old one, it doesn't pick a side, it writes both with dates, and from then on it treats you like someone who might be wrong the raw folder is a museum... the wiki is the actual working brain full breakdown is in the article below

kai

32,120 görüntüleme • 1 ay önce