Video wird geladen...

Video konnte nicht geladen werden

Beim Laden dieses Videos ist ein Problem aufgetreten. Dies könnte an einem vorübergehenden Netzwerkproblem liegen oder das Video ist möglicherweise nicht verfügbar.

I got Llama 3 running in my browser using only my GPU with my Wi-Fi switched OFF completely client-side WebGPU is a new feature in browsers where JS can use the GPU of the device and apparently you can run LLMs on it too, and it's fast! My end... goal for 🧠 is a locally run client side 100% private conversation where nothing gets sent to, processed in or stored in the cloud This will be helpful because I prefer making it web based instead of coding for 4 different native platforms (iOS, Android, MacOS, Windows) Two challenges still: 1) the user has to download the LLM model first and the smallest model is still ~3GB, but they only have to download it once 2) not all devices have fast GPUs yet for LLMs, but it's starting to become a default in most devices e.g.s smartphonesshow more

@levelsio

949,062 subscribers

261,455 Aufrufe • vor 2 Jahren •via X (Twitter)

Wissenschaft & Technologie

Anya Rossi• Live Now

Private livecam show

9 Kommentare

Profilbild von @levelsio

@levelsiovor 2 Jahren

One way to deal with the limitations of having to download model - let user use cloud model but show an indicator ☁️ Cloud Based Conversation or smth then once it's locally ran switch to ✅ 100% private local conversation - if user GPU or connection doesn't allow for locally run LLM give them the cloud one

Profilbild von Mario Hachemer

Mario Hachemervor 2 Jahren

I don't think users will be satisfied with the quality of even 7b models for that use case. Local compute is techie-brain thinking, no b2c product has ever won because of privacy advantages.

Profilbild von @levelsio

@levelsiovor 2 Jahren

Yep but there's 70B too, issue is how do you get that 40GB file to the user

Profilbild von @levelsio

@levelsiovor 2 Jahren

LM Studio is great but unusable for non-techies, it's still very hard You have to download it, install it, go to the right tab, search the model, choose the right quantization (this is insane btw, why doesn't it choose the best for me) It's very engineer minded and not user friendly

Profilbild von Xenova

Xenovavor 2 Jahren

While WebGPU-powered LLMs are amazing for demos (and to show what's possible in a browser context), I think the most exciting applications will use smaller task-specific models (especially for vision and multimodal tasks). For example: - Background removal: - Image segmentation (SAM):

Profilbild von Francisco Cordoba - Francisco Money

Francisco Cordoba - Francisco Moneyvor 2 Jahren

Unless your users are tech savy and super concern about privacy they are not going to download a file.

Profilbild von @levelsio

@levelsiovor 2 Jahren

It'll be automatic ofc

Profilbild von hugo alves

hugo alvesvor 2 Jahren

Still don’t understand why you named your product “the rapist ai”

Profilbild von Fleetwood

Fleetwoodvor 2 Jahren

On-device is 100% the future - particularly for private use cases like this. Combine this with TTS e.g so that users can speak & build the rapport. HF has been working hard on web + on-device with Transformers.JS & my project:

Ähnliche Videos

BOOMI MY NEW GROK DEVICE! I HAVE Grok ON AN OLD APPLE WATCH MADE IN TO A RETRO IPOD! Still working on it but my goal to have a remote single purpose Grok device has been achieved! With the benefits of posting my grifter slop on to X from this device also (that is another article). And a minimal phone. So how did I do it? First I got this IPod converter kit: it is a beautiful device to hold made of Aluminum. Second I got this app: “Ask Watch AI” it has a free version. Third I got my XAI Grok API Key: Toss a few bucks in the account to fund it. Finally I set up the system by answering questions. This is my first run. I am coding an app for the Apple Watch that will be easier than pressing “done”. I will write a full article on it as soon as I get it done. My goal is to use the cell plan on this Apple Watch and to have a minimal phone and Grok device. Soon these devices will be sent to Starlink. With full Grok Bot and Grok Build. No need for an OS or an App. Obviously this is an Apple Watch but soon it will be a device like this of course with a camera too.

BOOMI MY NEW GROK DEVICE! I HAVE Grok ON AN OLD APPLE WATCH MADE IN TO A RETRO IPOD! Still working on it but my goal to have a remote single purpose Grok device has been achieved! With the benefits of posting my grifter slop on to X from this device also (that is another article). And a minimal phone. So how did I do it? First I got this IPod converter kit: it is a beautiful device to hold made of Aluminum. Second I got this app: “Ask Watch AI” it has a free version. Third I got my XAI Grok API Key: Toss a few bucks in the account to fund it. Finally I set up the system by answering questions. This is my first run. I am coding an app for the Apple Watch that will be easier than pressing “done”. I will write a full article on it as soon as I get it done. My goal is to use the cell plan on this Apple Watch and to have a minimal phone and Grok device. Soon these devices will be sent to Starlink. With full Grok Bot and Grok Build. No need for an OS or an App. Obviously this is an Apple Watch but soon it will be a device like this of course with a camera too.

Brian Roemmele

33,846 Aufrufe • vor 22 Tagen

In LiveView, we keep the state in one place - the server - and it simplifies a lot. But sometimes the state on the client is preferred, or even necessary, and we can't use LiveView for that part. Unless we have a LiveView that runs on the client 👀 Using Popcorn, we managed to run LiveView fully in the browser – here's a POC 🎉 The approach is to have a fully client-side LiveView rendered by a server-side one, and they can communicate with each other. Try it out:

In LiveView, we keep the state in one place - the server - and it simplifies a lot. But sometimes the state on the client is preferred, or even necessary, and we can't use LiveView for that part. Unless we have a LiveView that runs on the client 👀 Using Popcorn, we managed to run LiveView fully in the browser – here's a POC 🎉 The approach is to have a fully client-side LiveView rendered by a server-side one, and they can communicate with each other. Try it out:

Elixir by Software Mansion

13,537 Aufrufe • vor 9 Monaten

A voice agent powered by gpt-oss. Running locally on my macBook. Demo recorded in a Waymo with WiFi turned off. I'm still on my space game voice AI kick, obviously. Code link below. For conversational voice AI, you want to set the gpt-oss reasoning behavior to "low". (The default is "medium".) Notes on how to do that and a jinja template you can use are in the repo. The LLM in the demo video is the big, 120B version of gpt-oss. You can use the smaller, 20B model for this, of course. But OpenAI really did a cool thing here designing the 120B model to run in "just" 80GB of VRAM. And the llama.cpp mlx inference is fast: ~250ms TTFT. Running a big model on-device feels like a time warp into the future of AI.

A voice agent powered by gpt-oss. Running locally on my macBook. Demo recorded in a Waymo with WiFi turned off. I'm still on my space game voice AI kick, obviously. Code link below. For conversational voice AI, you want to set the gpt-oss reasoning behavior to "low". (The default is "medium".) Notes on how to do that and a jinja template you can use are in the repo. The LLM in the demo video is the big, 120B version of gpt-oss. You can use the smaller, 20B model for this, of course. But OpenAI really did a cool thing here designing the 120B model to run in "just" 80GB of VRAM. And the llama.cpp mlx inference is fast: ~250ms TTFT. Running a big model on-device feels like a time warp into the future of AI.

kwindla

202,211 Aufrufe • vor 1 Jahr

turns out my habits were only half of it 9 steps to cut your Grok Bot usage by over 50% (if you're on a Cursor plan): 1. open cursor desktop app settings 2. go to the models option in the side menu 3. flip off any models you don't want to be using 4. next, set task models in the top of the page to a model you prefer that's cost efficient (do not leave it on default, this is IMPORTANT) 5. otherwise... your task models could auto set to high and/or fast... that's a more expensive model, and you get billed 2x 6. after, you'll click on the cloud agents option in your side menu, and it'll take you to the cursor web dashboard for your account 7. you will need to scroll down to the preferences section on that page and click on the default model option (again IMPORTANT) 8. set that to a cost efficient model you prefer and don't leave it on default... otherwise your cloud agents may use high and/or fast yet again 9. bonus tip (won't apply to most): lastly click on the spending option in your web dashboard's side menu and also make sure your on-demand usage is not on unlimited and is set to a cap you're comfortable with my token usage has reduced 60% after doing this and I'm sure it'll be different for everyone... but make sure to check this process can be a tad bit confusing too let me know if you need any help!

turns out my habits were only half of it 9 steps to cut your Grok Bot usage by over 50% (if you're on a Cursor plan): 1. open cursor desktop app settings 2. go to the models option in the side menu 3. flip off any models you don't want to be using 4. next, set task models in the top of the page to a model you prefer that's cost efficient (do not leave it on default, this is IMPORTANT) 5. otherwise... your task models could auto set to high and/or fast... that's a more expensive model, and you get billed 2x 6. after, you'll click on the cloud agents option in your side menu, and it'll take you to the cursor web dashboard for your account 7. you will need to scroll down to the preferences section on that page and click on the default model option (again IMPORTANT) 8. set that to a cost efficient model you prefer and don't leave it on default... otherwise your cloud agents may use high and/or fast yet again 9. bonus tip (won't apply to most): lastly click on the spending option in your web dashboard's side menu and also make sure your on-demand usage is not on unlimited and is set to a cap you're comfortable with my token usage has reduced 60% after doing this and I'm sure it'll be different for everyone... but make sure to check this process can be a tad bit confusing too let me know if you need any help!

klöss

35,455 Aufrufe • vor 12 Tagen

Lando: “In a way I wanted to keep number 4 because I'm just happy with that and it goes well with my logo. So that's how straightforward I am. But I will go with number 1. Because I have to. I have to. It could be my only opportunity in my life that I get to do such a thing. Not many people in the world get to do that kind of thing. And it's not for me. It's just for my mechanics. This is for my engineers. It's the fact that they get to run with the number 1 on the car. And I think it'll mean a lot more to them than it does to me as a number. And for me, that's a lot more important.”

naenia ¹ ⁶³

1,648,587 Aufrufe • vor 9 Monaten

This is the fastest I've seen Llama 3.3 running anywhere! Llama 3.3 70B running at 652 t/s is lightning fast. And if you want Llama 3.1, here are the speeds I was able to get: • Llama 3.1 8B: 1006 t/s • Llama 3.1 70B: 709 t/s • Llama 3.1 405B: 206 t/s (You can access all of these models for free! See the link below.) This speed is incredible, but the interesting part is what's happening behind the scenes: These models aren't running on a GPU! In this video, I'm using the SambaNova cloud to access these models. They built a custom chip (SN40L) optimized for AI workflows. A single SN40L chip can hold hundreds of models (trillions of parameters) in memory! The speed alone is a huge deal, but the big advantage is for agentic workflows running multiple specialized models. A GPU can only host a single model and switch (unload and load) to a different model if necessary. An SN40L, on the other hand, can host every model at once, making it much faster. Here is the video where you can see how fast these chips are:

This is the fastest I've seen Llama 3.3 running anywhere! Llama 3.3 70B running at 652 t/s is lightning fast. And if you want Llama 3.1, here are the speeds I was able to get: • Llama 3.1 8B: 1006 t/s • Llama 3.1 70B: 709 t/s • Llama 3.1 405B: 206 t/s (You can access all of these models for free! See the link below.) This speed is incredible, but the interesting part is what's happening behind the scenes: These models aren't running on a GPU! In this video, I'm using the SambaNova cloud to access these models. They built a custom chip (SN40L) optimized for AI workflows. A single SN40L chip can hold hundreds of models (trillions of parameters) in memory! The speed alone is a huge deal, but the big advantage is for agentic workflows running multiple specialized models. A GPU can only host a single model and switch (unload and load) to a different model if necessary. An SN40L, on the other hand, can host every model at once, making it much faster. Here is the video where you can see how fast these chips are:

Santiago

97,561 Aufrufe • vor 1 Jahr

working with cloud agents in cursor feels the same as local, but the fact that i can shut down my computer and still have it running is amazing. one of my favorite use cases is planning in cursor, then send off to cloud agent for impl. here's a short demo

working with cloud agents in cursor feels the same as local, but the fact that i can shut down my computer and still have it running is amazing. one of my favorite use cases is planning in cursor, then send off to cloud agent for impl. here's a short demo

eric zakariasson

61,966 Aufrufe • vor 10 Monaten

PowerInfer - a high-speed inference engine for deploying LLMs locally. Just came across this super interesting project on speeding up inference. It's not MoE but it's a simple approach that exploits the high locality in LLM inference to design a GPU-CPU hybrid inference engine. Hot-activated neurons are preloaded onto the GPU for fast access, while cold-activated neurons (the majority) are computed on the CPU. This approach significantly reduces GPU memory demands and CPU-GPU data transfer. It achieves an average token generation rate of 13.20 tokens/s, with a peak of 29.08 tokens/s, across various LLMs on a single NVIDIA RTX 4090 GPU. It's on only 18% lower than that achieved by a top-tier server-grade A100 GPU. It also significantly outperforms llama.cpp by up to 11.69x while retaining model accuracy. There is a lot more innovation around inference that's coming fast. Really encouraged by the study on sparse computation to enhance the computational efficiency of LLMs. It's now possible to use PowerInfer with Llama 2 and Faclon 40B. Mistral-7B support is coming soon!

PowerInfer - a high-speed inference engine for deploying LLMs locally. Just came across this super interesting project on speeding up inference. It's not MoE but it's a simple approach that exploits the high locality in LLM inference to design a GPU-CPU hybrid inference engine. Hot-activated neurons are preloaded onto the GPU for fast access, while cold-activated neurons (the majority) are computed on the CPU. This approach significantly reduces GPU memory demands and CPU-GPU data transfer. It achieves an average token generation rate of 13.20 tokens/s, with a peak of 29.08 tokens/s, across various LLMs on a single NVIDIA RTX 4090 GPU. It's on only 18% lower than that achieved by a top-tier server-grade A100 GPU. It also significantly outperforms llama.cpp by up to 11.69x while retaining model accuracy. There is a lot more innovation around inference that's coming fast. Really encouraged by the study on sparse computation to enhance the computational efficiency of LLMs. It's now possible to use PowerInfer with Llama 2 and Faclon 40B. Mistral-7B support is coming soon!

elvis

261,622 Aufrufe • vor 2 Jahren

Engineers, creatives, designers - lend me your ears, and eyes, to check my🥇1st tech demo of 2025: A #WebAI Agent running entirely client side in the browser, that's capable of controlling a flight search webpage, to get the job done! I'm using Google's Gemma 2 (2B) model in #JavaScript via #WebGPU thanks to the MediaPipe Web LLM library, combined with some extra function calling logic to enable advanced user experiences. If you like what you see, give this post a share, as I'm just one guy trying to show the world the power of Web AI to do great things. Try it for yourself using the link below after watching the video, but do so on a modernish machine (I have verified it runs on my 6 year old Dell XPS with only an Intel integrated GPU) but use a modern browser like Chrome as it needs WebGPU to run. Want me to make a demo for your industry? Tell me what you need / want to see next! Let me know what you think, and if you have any questions, then drop them in the comments below! YouTube Video Link:

Engineers, creatives, designers - lend me your ears, and eyes, to check my🥇1st tech demo of 2025: A #WebAI Agent running entirely client side in the browser, that's capable of controlling a flight search webpage, to get the job done! I'm using Google's Gemma 2 (2B) model in #JavaScript via #WebGPU thanks to the MediaPipe Web LLM library, combined with some extra function calling logic to enable advanced user experiences. If you like what you see, give this post a share, as I'm just one guy trying to show the world the power of Web AI to do great things. Try it for yourself using the link below after watching the video, but do so on a modernish machine (I have verified it runs on my 6 year old Dell XPS with only an Intel integrated GPU) but use a modern browser like Chrome as it needs WebGPU to run. Want me to make a demo for your industry? Tell me what you need / want to see next! Let me know what you think, and if you have any questions, then drop them in the comments below! YouTube Video Link:

Jason Mayes

42,895 Aufrufe • vor 1 Jahr

You can now try Llama 3.1 405B for free (link below)! This is the largest open-source model out there, and for the first time, an open model is competitive with closed models. This time around, Meta did something new: Llama 3.1 has a license that allows developers to use it to enhance other models. For the first time, you can distill Llama 3.1 405B's capabilities into a smaller, more practical model for your use case. First, here is the link where you can play with Llama 3.1 for free: The model is hosted in Tune Studio, an end-to-end platform for developing applications using Large Language Models. They are sponsoring this post. Take a look at the attached video. It will show you how you can fine-tune a simple model using Llama 3.1 without leaving the platform: 1. You can create an empty dataset 2. Use the playground to generate and record interactions with Llama 3.1 3. Modify the dataset directly using the playground 4. Export the data and fine-tune a smaller model Fast and easy! As long as you have a web browser, you can start experimenting with fine-tuning and Llama 3.1. That's all it takes!

You can now try Llama 3.1 405B for free (link below)! This is the largest open-source model out there, and for the first time, an open model is competitive with closed models. This time around, Meta did something new: Llama 3.1 has a license that allows developers to use it to enhance other models. For the first time, you can distill Llama 3.1 405B's capabilities into a smaller, more practical model for your use case. First, here is the link where you can play with Llama 3.1 for free: The model is hosted in Tune Studio, an end-to-end platform for developing applications using Large Language Models. They are sponsoring this post. Take a look at the attached video. It will show you how you can fine-tune a simple model using Llama 3.1 without leaving the platform: 1. You can create an empty dataset 2. Use the playground to generate and record interactions with Llama 3.1 3. Modify the dataset directly using the playground 4. Export the data and fine-tune a smaller model Fast and easy! As long as you have a web browser, you can start experimenting with fine-tuning and Llama 3.1. That's all it takes!

Santiago

55,609 Aufrufe • vor 2 Jahren

The latest MLX has a CUDA back-end! To get started: pip install "mlx[cuda]" With the same codebase you can develop locally, run your model on Apple silicon, or in the cloud on Nvidia GPUs. MLX is designed around Apple silicon - which has a unified memory architecture. It uses the same design with CUDA. So there's no need to move arrays around from CPU memory to GPU memory. Note, this is early days - some operations are missing and performance is still being optimized. But it's already quite fast for Transformer training, text generation, and more! Here's a demo using mlx-lm to generate text with Llama 3 8B (bf16) on an A100:

The latest MLX has a CUDA back-end! To get started: pip install "mlx[cuda]" With the same codebase you can develop locally, run your model on Apple silicon, or in the cloud on Nvidia GPUs. MLX is designed around Apple silicon - which has a unified memory architecture. It uses the same design with CUDA. So there's no need to move arrays around from CPU memory to GPU memory. Note, this is early days - some operations are missing and performance is still being optimized. But it's already quite fast for Transformer training, text generation, and more! Here's a demo using mlx-lm to generate text with Llama 3 8B (bf16) on an A100:

Awni Hannun

42,810 Aufrufe • vor 1 Jahr

notes on my reinforcement learning for rocket league side quest so far: i’m in awe that — we live in a time where i can train a reinforcement learning model that is learning to play rocket league and at the same time, test community made rocket league models. where, both processes can run in parallel & locally on my hardware. in the game window — i’m playing against a reinforcement learning ppo model that’s running locally. and in the vscode window — i’m training my own reinforcement learning model (just a dummy example script for now). also, the rocket league botting ecosystem has incredible tooling where you can test out different reinforcement learning ideas against other algorithms + it has great documentation for linux and windows. it’s honestly a beautiful hidden gem of a rabbit hole :)

notes on my reinforcement learning for rocket league side quest so far: i’m in awe that — we live in a time where i can train a reinforcement learning model that is learning to play rocket league and at the same time, test community made rocket league models. where, both processes can run in parallel & locally on my hardware. in the game window — i’m playing against a reinforcement learning ppo model that’s running locally. and in the vscode window — i’m training my own reinforcement learning model (just a dummy example script for now). also, the rocket league botting ecosystem has incredible tooling where you can test out different reinforcement learning ideas against other algorithms + it has great documentation for linux and windows. it’s honestly a beautiful hidden gem of a rabbit hole :)

naklecha

46,795 Aufrufe • vor 1 Jahr

Gavin Baker (Gavin Baker) says the disaggregation of inference can extend GPU useful lives from 3-4 years to 10-15. That may single-handedly save private credit and reduce the financing rates for GPUs, which will drive demand and help finance the build-out. "The disaggregation of prefill and inference is going to be amazing for the useful lives of GPU and may single-handedly save private credit. Private credit is in pain from these SaaS loans. But there's a lot of private credit in GPUs too. They were underwriting that to 3-4. The disaggregation of inference means that these GPUs are going to have 10 or 15-year lives. The AI skeptics are like, "Oh, these companies are all cooking their books. The useful life of a GPU is only a year or two. The useful life of a CPU is only four years because the rapid technological change." No. What rapid technological change has done with the disaggregation of prefill and inference is you can put a Cerebras system or Groq LPUs effectively in front of a Hopper or even an Ampere, use that Hopper and Ampere for prefill, and extend the useful life of that GPU until it melts. This is going to be really good for the whole private credit industry. It's gonna help finance the AI build-out. Because if you can start to finance GPUs at 5% or 6% instead of – I think CoreWeave's lowest financing was low sevens – that actually mathematically changes the cost to finance this build-out."

Gavin Baker (Gavin Baker) says the disaggregation of inference can extend GPU useful lives from 3-4 years to 10-15. That may single-handedly save private credit and reduce the financing rates for GPUs, which will drive demand and help finance the build-out. "The disaggregation of prefill and inference is going to be amazing for the useful lives of GPU and may single-handedly save private credit. Private credit is in pain from these SaaS loans. But there's a lot of private credit in GPUs too. They were underwriting that to 3-4. The disaggregation of inference means that these GPUs are going to have 10 or 15-year lives. The AI skeptics are like, "Oh, these companies are all cooking their books. The useful life of a GPU is only a year or two. The useful life of a CPU is only four years because the rapid technological change." No. What rapid technological change has done with the disaggregation of prefill and inference is you can put a Cerebras system or Groq LPUs effectively in front of a Hopper or even an Ampere, use that Hopper and Ampere for prefill, and extend the useful life of that GPU until it melts. This is going to be really good for the whole private credit industry. It's gonna help finance the AI build-out. Because if you can start to finance GPUs at 5% or 6% instead of – I think CoreWeave's lowest financing was low sevens – that actually mathematically changes the cost to finance this build-out."

Invest Like the Best

211,336 Aufrufe • vor 3 Monaten

Cursor now has a built-in web browser. You can say "start my app" and it will run the dev server in a terminal and then navigate the browser to localhost:3000. It can take screenshots of the page for iterating on design, and you can send DOM elements to the agent with prompts!

Cursor now has a built-in web browser. You can say "start my app" and it will run the dev server in a terminal and then navigate the browser to localhost:3000. It can take screenshots of the page for iterating on design, and you can send DOM elements to the agent with prompts!

Lee Robinson

216,252 Aufrufe • vor 10 Monaten

BREAKING: GPT-5.6 Sol is out—AND Codex has been merged into ChatGPT Desktop as ChatGPT Codex. This combo model and desktop app harness are the gold-standard for knowledge work in AI. 5.6 is powerful, fast, half the price of Fable, and my default for almost everything. We’ve been testing it internally Every 🪨 for about a month across coding, writing, design, and knowledge work. Here’s our day-zero vibe check: - An A-tier coder—but it’s not Fable. Sol scored 56/100 on our Senior Engineer benchmark compared to a 91 for Fable. I think the 56/100 undersells it, it's an excellent implementor, and very smart. But Fable just writes conceptually cleaner code and works better at the top end of task complexity. PRO-TIP: Use GPT-5.6 as Fable's subagent for the most goated combo in AI coding. - The best writer of the frontier models. It’s clearer and more concise than Fable or Opus 4.8, without the overexplaining or weird private language. It can one-shot marketing emails, help you workshop taglines, and explain complex concepts clearly. It's also super fast, which makes it easy to collaborate with. - Design is better, but not top-tier. It has noticeably more taste than 5.5, but Fable and Opus 4.8 are still playing at a different level. See examples in the video and vibe check below. - The real leap is knowledge work. Sol is the first model I’ve trusted to run whole loops of knowledge work—not just help with individual tasks. I use it to process email, surface decisions from meetings and Slack, find job candidates, scan Facebook Marketplace for furniture, and log my meals. It has shifted my job from doing the work to tending the system that does it. - The merged app is fine. I was extremely worried about this because I love the Codex app. OpenAI was caught in an interesting position: How to make an agent orchestration app for regular ChatGPT consumers, coders, and businesses all in one app. They now split the interface between ChatGPT Work and ChatGPT Codex. They're basically the same except Work hides code. And "Chat" has been demoted to 2nd tier status for quick questions in either one. It's not a big leap, but it's not a huge setback either. And it remains my favorite of the desktop agent orchestration apps. Verdict: If I really had to put my finger on it, I'd say Fable has way more big model smell. But that means it's a skill in itself to get value out of it—99% of people are still not there yet. GPT-5.6 is almost as powerful, but is easy to use, fast, and relatively cheap. It should give you an early sense of where model work is going. Full Every 🪨 Vibe Check:

BREAKING: GPT-5.6 Sol is out—AND Codex has been merged into ChatGPT Desktop as ChatGPT Codex. This combo model and desktop app harness are the gold-standard for knowledge work in AI. 5.6 is powerful, fast, half the price of Fable, and my default for almost everything. We’ve been testing it internally Every 🪨 for about a month across coding, writing, design, and knowledge work. Here’s our day-zero vibe check: - An A-tier coder—but it’s not Fable. Sol scored 56/100 on our Senior Engineer benchmark compared to a 91 for Fable. I think the 56/100 undersells it, it's an excellent implementor, and very smart. But Fable just writes conceptually cleaner code and works better at the top end of task complexity. PRO-TIP: Use GPT-5.6 as Fable's subagent for the most goated combo in AI coding. - The best writer of the frontier models. It’s clearer and more concise than Fable or Opus 4.8, without the overexplaining or weird private language. It can one-shot marketing emails, help you workshop taglines, and explain complex concepts clearly. It's also super fast, which makes it easy to collaborate with. - Design is better, but not top-tier. It has noticeably more taste than 5.5, but Fable and Opus 4.8 are still playing at a different level. See examples in the video and vibe check below. - The real leap is knowledge work. Sol is the first model I’ve trusted to run whole loops of knowledge work—not just help with individual tasks. I use it to process email, surface decisions from meetings and Slack, find job candidates, scan Facebook Marketplace for furniture, and log my meals. It has shifted my job from doing the work to tending the system that does it. - The merged app is fine. I was extremely worried about this because I love the Codex app. OpenAI was caught in an interesting position: How to make an agent orchestration app for regular ChatGPT consumers, coders, and businesses all in one app. They now split the interface between ChatGPT Work and ChatGPT Codex. They're basically the same except Work hides code. And "Chat" has been demoted to 2nd tier status for quick questions in either one. It's not a big leap, but it's not a huge setback either. And it remains my favorite of the desktop agent orchestration apps. Verdict: If I really had to put my finger on it, I'd say Fable has way more big model smell. But that means it's a skill in itself to get value out of it—99% of people are still not there yet. GPT-5.6 is almost as powerful, but is easy to use, fast, and relatively cheap. It should give you an early sense of where model work is going. Full Every 🪨 Vibe Check:

Dan Shipper

146,074 Aufrufe • vor 2 Monaten

React tip: "use client" misconceptions (1/5) 🚫 You need to mark EVERY individual file with "use client" to ensure it is a Client Component. ✅ You only need to mark each client entry point ONCE. Modules imported into a "use client" file are already part of the client bundle. (0:05) Adding "use client" to `3.js` marks the file as a client entry point, including it in the client bundle. (0:11) Importing `d()` into `3.js` (a client entry point) also includes it in the client bundle. (0:14) Adding "use client" to `4.js` has no additional bundling effect on `d()` because it is already part of the client bundle. (0:23) All modules (and their dependencies) imported to `3.js` are included in the client bundle. (0:26) Removing "use client" from `3.js` moves it and its dependencies into the server bundle.

React tip: "use client" misconceptions (1/5) 🚫 You need to mark EVERY individual file with "use client" to ensure it is a Client Component. ✅ You only need to mark each client entry point ONCE. Modules imported into a "use client" file are already part of the client bundle. (0:05) Adding "use client" to `3.js` marks the file as a client entry point, including it in the client bundle. (0:11) Importing `d()` into `3.js` (a client entry point) also includes it in the client bundle. (0:14) Adding "use client" to `4.js` has no additional bundling effect on `d()` because it is already part of the client bundle. (0:23) All modules (and their dependencies) imported to `3.js` are included in the client bundle. (0:26) Removing "use client" from `3.js` moves it and its dependencies into the server bundle.

Delba

157,404 Aufrufe • vor 2 Jahren

You don't need a PC to run Conso prompts to earn. You can do it directly from your phone. It may not be as fast as using a PC, and it can lag sometimes, but it works perfectly. Here is how: Step 1: Download the Misis browser. Click the link, download the extension, insert your invitation code, and sign in using your email. Step 2: Open another tab in the same Misis browser and sign in to ChatGPT. Open another tab for Gemini and do the same for your remaining AIs. That’s it. You’re ready to run your prompts to earn on CONSO from your phone.

You don't need a PC to run Conso prompts to earn. You can do it directly from your phone. It may not be as fast as using a PC, and it can lag sometimes, but it works perfectly. Here is how: Step 1: Download the Misis browser. Click the link, download the extension, insert your invitation code, and sign in using your email. Step 2: Open another tab in the same Misis browser and sign in to ChatGPT. Open another tab for Gemini and do the same for your remaining AIs. That’s it. You’re ready to run your prompts to earn on CONSO from your phone.

SCALE.

51,409 Aufrufe • vor 14 Tagen

David Friedberg: Frontier Models are Training on Your Novel Insights as “De-Identified Data” @jason: “Should they trust any of these LLMs with their proprietary knowledge for fear of having it cribbed into a core LLM?” david friedberg: “I have had experiences where we've asked some fairly novel scientific questions, and (the AI model) identifies it as a novel insight. It's like, ‘Oh, never thought about that, interesting, blah, blah, blah.’ And then using a different account, asking the next version (of the model) later, I've now experienced this. It's like, ‘Oh, well, you could do this,’ and it actually just describes this exact thing that we had in our chat in the previous version. Now, these are a handful of anecdotal experiences, but I know the domain that we work in, and the niche of it, and the ideation of this stuff, and the novelty of this stuff, and the lack of papers being published, and so on. So I know that there isn't some new corpus of information out there that's training the new model. So all I can say at that point is that my conversation or our analyses have been used for training.” David Sacks: “Okay, this does raise a really good question. What does it mean that the model is allowed to train on unidentifiable data?” Friedberg: “Well, that's my point. So it doesn't use any of my personal information, but it can use an insight derived from our chat, which it can then say is some training data that is unrelated. But the truth is, it's actually a piece of IP that's our organization’s IP, and our engagement back and forth. We don't have any NDA or confidentiality provisions or protections with them being a service provider back to us. This is why I care a lot about open source because I don't want them having my chat logs because they can use it for training to create an IP advantage that is now diffused to the rest of the market.”

David Friedberg: Frontier Models are Training on Your Novel Insights as “De-Identified Data” @jason: “Should they trust any of these LLMs with their proprietary knowledge for fear of having it cribbed into a core LLM?” david friedberg: “I have had experiences where we've asked some fairly novel scientific questions, and (the AI model) identifies it as a novel insight. It's like, ‘Oh, never thought about that, interesting, blah, blah, blah.’ And then using a different account, asking the next version (of the model) later, I've now experienced this. It's like, ‘Oh, well, you could do this,’ and it actually just describes this exact thing that we had in our chat in the previous version. Now, these are a handful of anecdotal experiences, but I know the domain that we work in, and the niche of it, and the ideation of this stuff, and the novelty of this stuff, and the lack of papers being published, and so on. So I know that there isn't some new corpus of information out there that's training the new model. So all I can say at that point is that my conversation or our analyses have been used for training.” David Sacks: “Okay, this does raise a really good question. What does it mean that the model is allowed to train on unidentifiable data?” Friedberg: “Well, that's my point. So it doesn't use any of my personal information, but it can use an insight derived from our chat, which it can then say is some training data that is unrelated. But the truth is, it's actually a piece of IP that's our organization’s IP, and our engagement back and forth. We don't have any NDA or confidentiality provisions or protections with them being a service provider back to us. This is why I care a lot about open source because I don't want them having my chat logs because they can use it for training to create an IP advantage that is now diffused to the rest of the market.”

The All-In Podcast

54,050 Aufrufe • vor 4 Tagen