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.

We now have local computer-use! M3 Pro 18GB running both UI-TARS-1.5-7B-6bit and a macOS sequoia VM entirely locally using MLX and c/ua at ~30second/action. Do it yourself here :

Dhruv

4,011 subscribers

76,871 views • 1 year ago •via X (Twitter)

Science & Technology Gaming Education

Anya Rossi• Live Now

Private livecam show

6 Comments

Luca Beurer-Kellner1 year ago

Nice, how is your experience with UI-TARS capability-wise? In our experiments it was a bit limited. Also, 30sec/action is wild. Looking for 30 actions/sec 😅

Rainmaker2 years ago

Here I share an XGBoost model that delivers a 25% CAGR with minimal drawdown on Visa stock. In this free Substack post I share code and commentary for a powerful Machine Learning strategy that delivers powerful returns.

coin☆witch1 year ago

giving my agent a mac hell yeah

Dami Dina1 year ago

@awnihannun How did you configure the MLX part?

EcoEconomist_DE1 year ago

The M3 Pro handling both UI-TARS and Sequoia VM locally shows Apple Silicon's growing AI muscle. c/ua's approach is particularly clever - curious about MLX optimization specifics though?

Fabrizio Milo1 year ago

What UI are you using?

Related Videos

🔴 The Pain: Running local MLX models is incredibly fast and private. But let's be real - testing tool calling via terminal is clunky, and there's zero good UI for it. 🟢 The Fix: rapid-mlx share Just ONE command gives you a polished web chat + seamless tool calling (works beautifully with gemma-4-12b-qat). We are proud to be the ONLY MLX inference engine in the community shipping this. ⚡️ 👇 Try it now: brew install raullenchai/rapid-mlx/rapid-mlx

🔴 The Pain: Running local MLX models is incredibly fast and private. But let's be real - testing tool calling via terminal is clunky, and there's zero good UI for it. 🟢 The Fix: rapid-mlx share Just ONE command gives you a polished web chat + seamless tool calling (works beautifully with gemma-4-12b-qat). We are proud to be the ONLY MLX inference engine in the community shipping this. ⚡️ 👇 Try it now: brew install raullenchai/rapid-mlx/rapid-mlx

raullen

22,760 views • 2 months ago

ChatGPT Voice is now in the desktop app. Control your computer and direct multiple agents running in ChatGPT Work or Codex, using just your voice. It's powered by GPT-Live, so it can speak, listen, and coordinate work in the app at the same time. Rolling out globally today on macOS and Windows to Plus, Pro, Business, Edu, and Enterprise plans.

ChatGPT Voice is now in the desktop app. Control your computer and direct multiple agents running in ChatGPT Work or Codex, using just your voice. It's powered by GPT-Live, so it can speak, listen, and coordinate work in the app at the same time. Rolling out globally today on macOS and Windows to Plus, Pro, Business, Edu, and Enterprise plans.

OpenAI

4,610,984 views • 17 days ago

Local Autonomous Agent for All of Us Browser-Use Web UI is a gradio app that uses LLMs to automatically surf the web to achieve tasks. All you need to do is just tell it what to do. It works like magic. And now anyone can run it locally with 1 click.

Local Autonomous Agent for All of Us Browser-Use Web UI is a gradio app that uses LLMs to automatically surf the web to achieve tasks. All you need to do is just tell it what to do. It works like magic. And now anyone can run it locally with 1 click.

cocktail peanut

40,782 views • 1 year ago

ParoQuant just got a big upgrade 🚀 ✅ Supports the new Qwen3.5 models ⚡ Now runs on MLX (fast local inference on Apple Silicon) 🧠 Preserves reasoning quality with 4-bit quantization We also built an agent demo running locally on my 4-year-old M2 Max. Can't wait to upgrade to an M5 Max and see what kind of magic we can do. ✨

ParoQuant just got a big upgrade 🚀 ✅ Supports the new Qwen3.5 models ⚡ Now runs on MLX (fast local inference on Apple Silicon) 🧠 Preserves reasoning quality with 4-bit quantization We also built an agent demo running locally on my 4-year-old M2 Max. Can't wait to upgrade to an M5 Max and see what kind of magic we can do. ✨

Zhijian Liu

49,442 views • 5 months ago

Hey Theo - t3.gg my Tesla has a local model running it. It works without being connected at all. And doesn’t kill me. Just got home. Alibaba Cloud researchers told me this week at ACL that the real action in the labs is building smaller models. So the open source & “run locally” bulls will have their day.

Hey Theo - t3.gg my Tesla has a local model running it. It works without being connected at all. And doesn’t kill me. Just got home. Alibaba Cloud researchers told me this week at ACL that the real action in the labs is building smaller models. So the open source & “run locally” bulls will have their day.

Robert Scoble

18,428 views • 1 month ago

oMLX last version seems having some issues in decoding. 🧐 Here a test with Qwen3.6-35B-A3B-MLX-6bit on M5 Max using. 🥇 LM Studio MLX 1.8.5 → 100.9 toks/s 🥈 mlx-vlm 0.6.2 → 100.1 toks/s 🥉 oMLX 0.4.2 dev3 → 58.7 toks/s 👀 Avg Gen TPS: oMLX 58.7 → LM Studio 100.9 (+71.8%) I have to thank pymike00 that raised this oMLX issue after seeing my video on using it with Codex. I bet there is a bug in oMLX chat and server at the moment, because internal benchmarks are ok, video attached. I bet Jun Kim will fix this soon 💪

oMLX last version seems having some issues in decoding. 🧐 Here a test with Qwen3.6-35B-A3B-MLX-6bit on M5 Max using. 🥇 LM Studio MLX 1.8.5 → 100.9 toks/s 🥈 mlx-vlm 0.6.2 → 100.1 toks/s 🥉 oMLX 0.4.2 dev3 → 58.7 toks/s 👀 Avg Gen TPS: oMLX 58.7 → LM Studio 100.9 (+71.8%) I have to thank pymike00 that raised this oMLX issue after seeing my video on using it with Codex. I bet there is a bug in oMLX chat and server at the moment, because internal benchmarks are ok, video attached. I bet Jun Kim will fix this soon 💪

Ivan Fioravanti ᯅ

20,066 views • 2 months ago

WOW! I'm so excited about this. OpenAI Developers said Codex was good at Computer Use, but I wasn't prepared for this. For the last two weeks I've been working on a Computer Use skill to work with Linux. And while I had some success, it was a pretty frustrating experience. That is...until the breakthrough. Using accessibility tools, Codex can now control my entire computer, not just the browser. There are limits to this, of course, but what a time to be alive. This Computer Use skill will unlock and entirely new set of automations, all powered by Codex. Demonstration below 👇

WOW! I'm so excited about this. OpenAI Developers said Codex was good at Computer Use, but I wasn't prepared for this. For the last two weeks I've been working on a Computer Use skill to work with Linux. And while I had some success, it was a pretty frustrating experience. That is...until the breakthrough. Using accessibility tools, Codex can now control my entire computer, not just the browser. There are limits to this, of course, but what a time to be alive. This Computer Use skill will unlock and entirely new set of automations, all powered by Codex. Demonstration below 👇

am.will

79,657 views • 4 months ago

announcing evo x Claude Science you can now run EVO natively inside claude science with just a few clicks. Use the data integrations and form factor of Claude science to now use EVO to run long running autoresearch runs inside it- setting up baselines, running hundreds of gated experiments, keep a durable scored frontier, hill-climb unattended. On the compute you already use. Best of both worlds. we have over 25k+ users and a wonderful community of bio researchers that use EVO. Cant wait to see what all you do with it !

announcing evo x Claude Science you can now run EVO natively inside claude science with just a few clicks. Use the data integrations and form factor of Claude science to now use EVO to run long running autoresearch runs inside it- setting up baselines, running hundreds of gated experiments, keep a durable scored frontier, hill-climb unattended. On the compute you already use. Best of both worlds. we have over 25k+ users and a wonderful community of bio researchers that use EVO. Cant wait to see what all you do with it !

Alok Bishoyi

12,875 views • 1 month ago

Next mlx-vlm release will ship with continuous batching support on the server 🚀 What's coming: → Continuous batching — new requests join the active batch immediately, no waiting. Mixed image + text batches supported → OpenAI-compatible API — field-for-field match with mlx-lm, reasoning/content split for thinking models, tag-aware streaming → Multi-turn tool calling — full tool use support across streaming and non-streaming, works with Gemma4 and other templates → Vision feature caching — cache image embeddings across turns. Gemma4: 228x speedup, Qwen3.5: 23x on cache hit All running locally on Apple Silicon. Check our this demo running 4 concurrent requests (mixed image + text) to gemma-4-26B-A4B-IT by Google Gemma in bf16 using Pi + MLX-VLM server on my M3 Ultra. One of the requests ingests a 8K resolution image!

Next mlx-vlm release will ship with continuous batching support on the server 🚀 What's coming: → Continuous batching — new requests join the active batch immediately, no waiting. Mixed image + text batches supported → OpenAI-compatible API — field-for-field match with mlx-lm, reasoning/content split for thinking models, tag-aware streaming → Multi-turn tool calling — full tool use support across streaming and non-streaming, works with Gemma4 and other templates → Vision feature caching — cache image embeddings across turns. Gemma4: 228x speedup, Qwen3.5: 23x on cache hit All running locally on Apple Silicon. Check our this demo running 4 concurrent requests (mixed image + text) to gemma-4-26B-A4B-IT by Google Gemma in bf16 using Pi + MLX-VLM server on my M3 Ultra. One of the requests ingests a 8K resolution image!

Prince Canuma

82,169 views • 3 months ago

.Sen. Maggie Hassan: "When I had a miscarriage at 12 weeks it was painful. It was hard and it was emotional and Mifepristone had nothing to do with it. I didn't use Mifepristone. I had to have a D&C. So the gaslighting here is extraordinary."

.Sen. Maggie Hassan: "When I had a miscarriage at 12 weeks it was painful. It was hard and it was emotional and Mifepristone had nothing to do with it. I didn't use Mifepristone. I had to have a D&C. So the gaslighting here is extraordinary."

CSPAN

375,225 views • 6 months ago

Claude Code is now included in the Claude Pro plan which means Pro plan subscribers can use their rate limits for Claude apps and Claude Code! To celebrate, here's a conversation with Claude Code's founding engineer, Boris Cherny, on: - how we use Claude Code at Anthropic - underrated tips for using it effectively - the future of coding tools - and more!

Claude Code is now included in the Claude Pro plan which means Pro plan subscribers can use their rate limits for Claude apps and Claude Code! To celebrate, here's a conversation with Claude Code's founding engineer, Boris Cherny, on: - how we use Claude Code at Anthropic - underrated tips for using it effectively - the future of coding tools - and more!

Alex Albert

123,756 views • 1 year ago

local LLMs on Mac are getting good. but the workflow around models is still messy. you have models in: •⁠ ⁠LM Studio •⁠ ⁠Ollama •⁠ ⁠Hugging Face cache •⁠ ⁠random folders •⁠ ⁠manual downloads and every runtime expects something slightly different. we built ModelHub to clean up this layer. it’s a native macOS menu bar app for discovering, downloading, and managing local LLMs. not a replacement for Ollama, oMLX or LM Studio. it sits beside them. Ollama = run models MLX = Apple Silicon acceleration LM Studio = local model workspace Hugging Face = model ecosystem ModelHub = model discovery + management layer we’re live on Product Hunt today. would love feedback from people running local models on Mac.

local LLMs on Mac are getting good. but the workflow around models is still messy. you have models in: •⁠ ⁠LM Studio •⁠ ⁠Ollama •⁠ ⁠Hugging Face cache •⁠ ⁠random folders •⁠ ⁠manual downloads and every runtime expects something slightly different. we built ModelHub to clean up this layer. it’s a native macOS menu bar app for discovering, downloading, and managing local LLMs. not a replacement for Ollama, oMLX or LM Studio. it sits beside them. Ollama = run models MLX = Apple Silicon acceleration LM Studio = local model workspace Hugging Face = model ecosystem ModelHub = model discovery + management layer we’re live on Product Hunt today. would love feedback from people running local models on Mac.

sabesh 📟

31,497 views • 2 months ago

The Quick desktop app is here, and it’s compelling. Connects to your email, calendar, Slack, local files, and several other apps to flag important communications, retrieve and summarize info, make recommendations, send communications, and create agents that do work you used to have to do yourself. Gets smarter and more personalized the more you use it. Been using it a lot recently and is changing how I work. It’s allowing me to use applications like my inbox more like an archive, and Quick as my personalized, prioritized, productivity hub that can multi-task various needs. Still early days, and a lot more coming, but excited for folks to start using it to make the undifferentiated work so much less complicated.

The Quick desktop app is here, and it’s compelling. Connects to your email, calendar, Slack, local files, and several other apps to flag important communications, retrieve and summarize info, make recommendations, send communications, and create agents that do work you used to have to do yourself. Gets smarter and more personalized the more you use it. Been using it a lot recently and is changing how I work. It’s allowing me to use applications like my inbox more like an archive, and Quick as my personalized, prioritized, productivity hub that can multi-task various needs. Still early days, and a lot more coming, but excited for folks to start using it to make the undifferentiated work so much less complicated.

Andy Jassy

270,853 views • 3 months ago

🚨Hurricane Helene victim SLAMS the local and federal government. “It’s weird to me that some 50-some-odd-year-old guy, in the middle of a disaster, has to reach out to the American government to explain to them, ‘Hey, you have resources. You have people who want to help, and you really do have the capacity to do it. Uncle Sam, you can do it.’ This is despicable. Joe, Kamala, do you know you have a military? Do you know that we have spent trillions of dollars on it, and that we can, at the drop of a hat, invade any country in the world, flatten them, and rebuild them? We’ve done it lots of times. Read a history book. Why can’t you use those same resources that we’ve already paid for, here, now? Where are you? I understand there was a vacation you needed to take, Joe, and Kamala, you had some fundraising for your campaign. Suspend that shit and get in here!”

🚨Hurricane Helene victim SLAMS the local and federal government. “It’s weird to me that some 50-some-odd-year-old guy, in the middle of a disaster, has to reach out to the American government to explain to them, ‘Hey, you have resources. You have people who want to help, and you really do have the capacity to do it. Uncle Sam, you can do it.’ This is despicable. Joe, Kamala, do you know you have a military? Do you know that we have spent trillions of dollars on it, and that we can, at the drop of a hat, invade any country in the world, flatten them, and rebuild them? We’ve done it lots of times. Read a history book. Why can’t you use those same resources that we’ve already paid for, here, now? Where are you? I understand there was a vacation you needed to take, Joe, and Kamala, you had some fundraising for your campaign. Suspend that shit and get in here!”

I Meme Therefore I Am 🇺🇸

146,561 views • 1 year ago

"It is honestly a little pathetic that our ancestors endured and fought in violent, bloody, years long wars and all we have to do is use our voices and risk being called mean names, and most of us can't even do that." "They fought in a bloody conflict just so they didn't have to be ruled over by a British King." "Here in modern day America we are faced with a future that could look vastly different than the culture we currently have." We are slowly allowing the American way of live to be changed and taken from us from enemies both foreign and domestic. Do something about it!

"It is honestly a little pathetic that our ancestors endured and fought in violent, bloody, years long wars and all we have to do is use our voices and risk being called mean names, and most of us can't even do that." "They fought in a bloody conflict just so they didn't have to be ruled over by a British King." "Here in modern day America we are faced with a future that could look vastly different than the culture we currently have." We are slowly allowing the American way of live to be changed and taken from us from enemies both foreign and domestic. Do something about it!

AmericanPapaBear™

16,387 views • 7 months ago

Leo Woodall says he and Havana Rose Liu undertook months of intensive piano training for 'TUNER': "I wasn't experienced at all with piano before the movie. We had months of training, both Havana and myself, hours a day, every day. It was wonderful. It was challenging, and I think we were both quite stressed that we didn't have enough time, but we just worked really hard at it to be able to do a convincing job." 'TUNER' is now playing in theaters.

Leo Woodall says he and Havana Rose Liu undertook months of intensive piano training for 'TUNER': "I wasn't experienced at all with piano before the movie. We had months of training, both Havana and myself, hours a day, every day. It was wonderful. It was challenging, and I think we were both quite stressed that we didn't have enough time, but we just worked really hard at it to be able to do a convincing job." 'TUNER' is now playing in theaters.

Film Updates

25,822 views • 2 months ago

You really can just do things! Use *any* Hugging Face space as a MCP server along with your Local Models! 🔥 Here in we use Qwen 3 30B A3B with ggml llama.cpp and Hugging Face tiny agents to create images via FLUX powered by ZeroGPU ⚡ It's quite a bit crazy to see local models be capable of so much and just be able to understand/ infer from tool description! There's a lot of potential here in automating video generation workflows, content curation and a lot more.. Bonus: you can plug any other Inference Provider if you don't want to run locally too! npx @ huggingface/tiny-agents run [TASK] oh, and we provide both typescript and python client! 🐐

You really can just do things! Use any Hugging Face space as a MCP server along with your Local Models! 🔥 Here in we use Qwen 3 30B A3B with ggml llama.cpp and Hugging Face tiny agents to create images via FLUX powered by ZeroGPU ⚡ It's quite a bit crazy to see local models be capable of so much and just be able to understand/ infer from tool description! There's a lot of potential here in automating video generation workflows, content curation and a lot more.. Bonus: you can plug any other Inference Provider if you don't want to run locally too! npx @ huggingface/tiny-agents run [TASK] oh, and we provide both typescript and python client! 🐐

Vaibhav (VB) Srivastav

39,113 views • 1 year ago

I'm blown away. Ornith-1.0 35b is the best local model I've ever run that doesn't require 200GB+ of RAM Incredible at coding According to my tests it's both smarter and faster than Qwen 3.6 Plus it runs like a dream on a DGX Spark Here's how I'm using it: I have it running every hour and looping through different parts of my Henry Intelligent Machines codebase It looks for security vulnerabilities Any vulnerabilities it finds, it writes a report with the fix Then once a day another loop I have running in Codex reads the report with all the security bugs and verifies, then fixes I also have it once a day going through a reviewing all my open PRs Local models running on loops 24/7 is the future. I promise you What's becoming possible today with local models is truly mind blowing

I'm blown away. Ornith-1.0 35b is the best local model I've ever run that doesn't require 200GB+ of RAM Incredible at coding According to my tests it's both smarter and faster than Qwen 3.6 Plus it runs like a dream on a DGX Spark Here's how I'm using it: I have it running every hour and looping through different parts of my Henry Intelligent Machines codebase It looks for security vulnerabilities Any vulnerabilities it finds, it writes a report with the fix Then once a day another loop I have running in Codex reads the report with all the security bugs and verifies, then fixes I also have it once a day going through a reviewing all my open PRs Local models running on loops 24/7 is the future. I promise you What's becoming possible today with local models is truly mind blowing

Alex Finn

314,695 views • 1 month ago

Raindrop cofounder ben hylak goes deeper on his exchange with Brian Chesky about booking Airbnbs with LLMs: "If Airbnb had an API, I would use it, and I would book Airbnbs through it using Claude Code. I know I would do it." "You see companies almost reducing themselves into an API with absolutely no moat. You look at Photoshop, Illustrator, etc. They're like, 'We have a Claude Code integration now.'" "At the point where people are just using Photoshop, Illustrator, etc. as an MCP, they've sort of lost the game, if no one is touching the UI anymore." "Right now, companies have to do that, increasingly, because they have no other choice. I think there will be a point where the incentives don't make sense anymore."

Raindrop cofounder ben hylak goes deeper on his exchange with Brian Chesky about booking Airbnbs with LLMs: "If Airbnb had an API, I would use it, and I would book Airbnbs through it using Claude Code. I know I would do it." "You see companies almost reducing themselves into an API with absolutely no moat. You look at Photoshop, Illustrator, etc. They're like, 'We have a Claude Code integration now.'" "At the point where people are just using Photoshop, Illustrator, etc. as an MCP, they've sort of lost the game, if no one is touching the UI anymore." "Right now, companies have to do that, increasingly, because they have no other choice. I think there will be a point where the incentives don't make sense anymore."

TBPN

109,350 views • 2 months ago