正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

Local AI: I can process complex logic, reasoning, and data. 🧠 Me: Cool, but can you draw an octopus? 🐙 Bonsai 2 Hadamard 27B: Hold my beer. 🍺 Watch it flex its multimodal muscles running locally on Rapid-MLX. Wait until the end for the final masterpiece! 🎨

Rapid-MLX · local AI on your Mac

13,837 subscribers

13,465 次观看 • 17 天前 •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

26 条评论

TVbat 的头像

TVbat17 天前

what the fuck is this song 😭

Chen 的头像

Chen17 天前

yeah. i keep underestimating how long the second half takes

JQ 的头像

JQ17 天前

What's the prefill and TPS?

Cathy@tknet.ai 的头像

[email protected]17 天前

Local multimodal on Mac is such a great direction.

Oga Focus 的头像

Oga Focus14 天前

The “hold my beer” moment was real

Omadudujoel 的头像

Omadudujoel14 天前

The octopus test might be the most important benchmark 😂

Gemba Masfd 的头像

Gemba Masfd14 天前

From reasoning to drawing in the same local model

Ohamara Kelechi 的头像

Ohamara Kelechi14 天前

Bonsai 2 flexing where it matters 🎨

Aminat Adesola02 的头像

Aminat Adesola0214 天前

Bonsai 2 has some serious visual capabilities

David Williams 的头像

David Williams14 天前

Would be interesting to see how it handles more complex visual instructions

miracle kefas 的头像

miracle kefas14 天前

That octopus came out better than expected 👀

Stephen Ufedo 的头像

Stephen Ufedo14 天前

Bonsai 2 really understood the assignment 🐙

adejare adefemi 的头像

adejare adefemi14 天前

27B on a local Mac doing multimodal work feels futuristic

John Smith 的头像

John Smith14 天前

This is the kind of local AI demo people actually remember

Kraven 的头像

Kraven14 天前

Running it locally makes the demo even more impressive 🔥

abdulrahman 的头像

abdulrahman14 天前

Complex reasoning is cool, but multimodal creativity hits different

Salam Adebowale Saheed 的头像

Salam Adebowale Saheed14 天前

That final drawing was the perfect payoff

Joy 的头像

Joy14 天前

The octopus is officially the new AI benchmark 🐙

ATM 的头像

ATM14 天前

27B doing multimodal work locally is seriously impressive

Wasiu7642 的头像

Wasiu764214 天前

Would be interesting to see how it handles more complex visual instructions

Latman 的头像

Latman14 天前

The octopus understood the assignment perfectly 🐙

RAIDEN 的头像

RAIDEN14 天前

Bonsai 2 has some serious visual capabilities

PROSPER GREGORY 的头像

PROSPER GREGORY14 天前

Vision plus generation opens up a completely different workflow

Pst Livingstone Bright 的头像

Pst Livingstone Bright13 天前

Imagine combining this with local tool calling and agents 🤖

Chika Benson Ejiofor 的头像

Chika Benson Ejiofor14 天前

27B doing multimodal work locally is seriously impressive

Jesuniyi Joshua 的头像

Jesuniyi Joshua14 天前

Hmm 🧐

相关视频

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

49,826 次观看 • 7 个月前

just built a local AI waifu on my two macbooks by using the oss parallax inference library!!! not super smooth yet but parallax is still the easiest way to run llms locally on top of mlx - and no bigco can steal my waifu. still need folks to help her see/hear/draw since mlx has no legit wrapper yet. i’m running a campaign giving away mac minis + dgx spark. please please share your localhost ai model setup by using parallax (you don't need a gpu to try self-hosting you can just start with an extra macbook you borrowed from your friend like i did) or any fun app built on top of yourself host model to link below. come build sovereign intelligence with me 🫶

just built a local AI waifu on my two macbooks by using the oss parallax inference library!!! not super smooth yet but parallax is still the easiest way to run llms locally on top of mlx - and no bigco can steal my waifu. still need folks to help her see/hear/draw since mlx has no legit wrapper yet. i’m running a campaign giving away mac minis + dgx spark. please please share your localhost ai model setup by using parallax (you don't need a gpu to try self-hosting you can just start with an extra macbook you borrowed from your friend like i did) or any fun app built on top of yourself host model to link below. come build sovereign intelligence with me 🫶

Asuka Zheng🎀@ICML

178,998 次观看 • 10 个月前

Superior methodology beats raw parameter count. 🧠 Introducing QVAC MedPsy: Local-first medical AI that redefines the possible. 1/ Unprecedented Power: MedPsy 1.7B model outperforms Google’s MedGemma 4B by 11 points and our 4B model beats MedGemma 27B on real-world health benchmarks. 2/ Extreme Efficiency: 3.2x fewer tokens means near-instant inference on your phone or wearable. 3/ Absolute Privacy: Expert-level reasoning running 100% locally. No data leaves your device. We aren’t simply shrinking models; we’re anchoring intelligence where it matters most. High-level medical logic is now a sovereign right. The future of healthcare is local. Learn more:

Superior methodology beats raw parameter count. 🧠 Introducing QVAC MedPsy: Local-first medical AI that redefines the possible. 1/ Unprecedented Power: MedPsy 1.7B model outperforms Google’s MedGemma 4B by 11 points and our 4B model beats MedGemma 27B on real-world health benchmarks. 2/ Extreme Efficiency: 3.2x fewer tokens means near-instant inference on your phone or wearable. 3/ Absolute Privacy: Expert-level reasoning running 100% locally. No data leaves your device. We aren’t simply shrinking models; we’re anchoring intelligence where it matters most. High-level medical logic is now a sovereign right. The future of healthcare is local. Learn more:

QVAC

2,416,342 次观看 • 5 个月前

This is how you build the next generation of AI applications. You need an idea, a good model architecture, and a great harness. That's what MUZIM did. It's an app that runs local multimodal and vision models offline to parse and index your images, videos, and documents frame by frame. • Super fast on-device processing • On-device video & image understanding • You can also bring your API keys for heavy reasoning • Your data stays where it is / nothing goes to the cloud You can do a couple of things with this: 1. Vibe searching for whatever you remember in photos or raw video clips. The app will jump straight to the exact frame and timestamp of the video. This is search as it should be. 2. Use local-running agents to perform tasks on your files. I show a couple of examples in my video. You can check it out here: Thanks to the MUZIM by OpenSoul team for showing me the tool and partnering with me on this post.

This is how you build the next generation of AI applications. You need an idea, a good model architecture, and a great harness. That's what MUZIM did. It's an app that runs local multimodal and vision models offline to parse and index your images, videos, and documents frame by frame. • Super fast on-device processing • On-device video & image understanding • You can also bring your API keys for heavy reasoning • Your data stays where it is / nothing goes to the cloud You can do a couple of things with this: 1. Vibe searching for whatever you remember in photos or raw video clips. The app will jump straight to the exact frame and timestamp of the video. This is search as it should be. 2. Use local-running agents to perform tasks on your files. I show a couple of examples in my video. You can check it out here: Thanks to the MUZIM by OpenSoul team for showing me the tool and partnering with me on this post.

Santiago

65,953 次观看 • 1 个月前

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 次观看 • 1 年前

MCP is mind-blowing. Here is a genius setup that will earn you a promotion with your manager. I have multiple data sources: data in S3, MySQL, and local files. Using Spice (an open-source project), I can co-locate all of that data behind a single SQL endpoint and augment it using MCP. Before, accessing data was slow, and I had to make multiple requests to get anything. After using Spice, I can query the data much faster (since it's mirrored locally), and I don't have to worry about where the data is coming from. Spice acts as an MCP server, so I can connect to it from my IDE or the Claude Desktop app and have access to all the data. But even more cool than that: Spice also acts as an MCP client, so it can connect to other MCP servers. This is where things get really interesting! Watch the video. I'm querying multiple data sources from my IDE and using the Google Maps MCP server for driving directions.

MCP is mind-blowing. Here is a genius setup that will earn you a promotion with your manager. I have multiple data sources: data in S3, MySQL, and local files. Using Spice (an open-source project), I can co-locate all of that data behind a single SQL endpoint and augment it using MCP. Before, accessing data was slow, and I had to make multiple requests to get anything. After using Spice, I can query the data much faster (since it's mirrored locally), and I don't have to worry about where the data is coming from. Spice acts as an MCP server, so I can connect to it from my IDE or the Claude Desktop app and have access to all the data. But even more cool than that: Spice also acts as an MCP client, so it can connect to other MCP servers. This is where things get really interesting! Watch the video. I'm querying multiple data sources from my IDE and using the Google Maps MCP server for driving directions.

Santiago

142,696 次观看 • 1 年前

AI has become a useful research partner. It can find information, summarize evidence, and suggest ideas. But can it help us think better? Can it help us navigate the complex nuances of high-stakes decisions? Today, we’re taking the next step toward our mission of scaling up good reasoning for high-stakes decisions with the launch of the Elicit Research Agent, an AI environment built for high-stakes decisions. You can use the Research Agent today at or access it through API ( to bring its capabilities into your own workflows.

AI has become a useful research partner. It can find information, summarize evidence, and suggest ideas. But can it help us think better? Can it help us navigate the complex nuances of high-stakes decisions? Today, we’re taking the next step toward our mission of scaling up good reasoning for high-stakes decisions with the launch of the Elicit Research Agent, an AI environment built for high-stakes decisions. You can use the Research Agent today at or access it through API ( to bring its capabilities into your own workflows.

Elicit

10,341 次观看 • 2 个月前

So I created a simple Expo app for remotely connecting to an OpenCode server (running on my mac) so I can remotely control & prompt from my iPad or phone. Not very polished but it works pretty well and its kinda interesting that you can see the feedback loop live.

So I created a simple Expo app for remotely connecting to an OpenCode server (running on my mac) so I can remotely control & prompt from my iPad or phone. Not very polished but it works pretty well and its kinda interesting that you can see the feedback loop live.

ryan vogel

50,706 次观看 • 9 个月前

What if your computer didn’t just sit there waiting for you? warmwind is turning the browser into an always-on AI work environment. You show it a workflow once → it can keep running it in the cloud. For example: SAP → check data → open email → process information → update the workflow. No complicated integrations. Warmwind can visually interact with the software clicking, typing, reading, and navigating screens like a human. You teach the workflow. It keeps doing the work. Watch the official launch video: Try Warmwind:

What if your computer didn’t just sit there waiting for you? warmwind is turning the browser into an always-on AI work environment. You show it a workflow once → it can keep running it in the cloud. For example: SAP → check data → open email → process information → update the workflow. No complicated integrations. Warmwind can visually interact with the software clicking, typing, reading, and navigating screens like a human. You teach the workflow. It keeps doing the work. Watch the official launch video: Try Warmwind:

AI Club

10,479 次观看 • 1 个月前

On a previous #TeawithGaryVee I was asked "What are the two most important traits you look for in an employee?" If you have a question for me, drop it here so my team can add it to the next Rapid Fire Q&A and follow me on

On a previous #TeawithGaryVee I was asked "What are the two most important traits you look for in an employee?" If you have a question for me, drop it here so my team can add it to the next Rapid Fire Q&A and follow me on

Gary Vaynerchuk

14,435 次观看 • 8 个月前

Okay, this is probably the best RAG UI I have ever seen. Just wow. I got a local version running in about 30 seconds with my existing Weaviate data. Here are some of the things I loved right away: 1) You can examine the raw data by collection 2) You can chat 3) You can see a tree for how the responses were generated 4) You can provide real-time feedback that can be used to train a model to improve the RAG workflow. 5) Still testing, but lots more! Well done Weaviate AI Database team!

Okay, this is probably the best RAG UI I have ever seen. Just wow. I got a local version running in about 30 seconds with my existing Weaviate data. Here are some of the things I loved right away: 1) You can examine the raw data by collection 2) You can chat 3) You can see a tree for how the responses were generated 4) You can provide real-time feedback that can be used to train a model to improve the RAG workflow. 5) Still testing, but lots more! Well done Weaviate AI Database team!

William J.B. Mattingly

33,340 次观看 • 1 年前

I guess Scott Wiener's talking points on regulating AI and tech are waiting on approval from the AI lobby. Until then, you can read my stance in Mission Local or check out our comprehensive plan to control AI so it works for humanity and not the oligarchs on our website.

I guess Scott Wiener's talking points on regulating AI and tech are waiting on approval from the AI lobby. Until then, you can read my stance in Mission Local or check out our comprehensive plan to control AI so it works for humanity and not the oligarchs on our website.

Saikat Chakrabarti

11,385 次观看 • 5 个月前

🔥Burn Your Desire is officially OUT NOW🔥 You can find my new Single on every streaming service and watch the Musicvideo right here: 🤘🏻 An absolute masterpiece if you ask me... I hope you can enjoy it too❤️‍🔥 #music #metal #metalcore #outnow

🔥Burn Your Desire is officially OUT NOW🔥 You can find my new Single on every streaming service and watch the Musicvideo right here: 🤘🏻 An absolute masterpiece if you ask me... I hope you can enjoy it too❤️‍🔥 #music #metal #metalcore #outnow

Ifas_core🏳️‍🌈

14,278 次观看 • 2 年前

On a previous #TeawithGaryVee I was asked "What do you think will be the highest growing asset someone can own digitally right now, especially with all the AI and web 2.0 buzz happening?" if you have a question for me, drop it here so my team can add it to the next Rapid Fire Q&A and follow me on

On a previous #TeawithGaryVee I was asked "What do you think will be the highest growing asset someone can own digitally right now, especially with all the AI and web 2.0 buzz happening?" if you have a question for me, drop it here so my team can add it to the next Rapid Fire Q&A and follow me on

Gary Vaynerchuk

24,240 次观看 • 2 个月前

On a previous #TeawithGaryVee I was asked "I'm a 70 year old woman and I wanna make some money for my retirement but I don't know where to start and what I can do to start something on social media. What advice do you have for me?" If you have a question for me, drop it here so my team can add it to the next Rapid Fire Q&A and follow me on

On a previous #TeawithGaryVee I was asked "I'm a 70 year old woman and I wanna make some money for my retirement but I don't know where to start and what I can do to start something on social media. What advice do you have for me?" If you have a question for me, drop it here so my team can add it to the next Rapid Fire Q&A and follow me on

Gary Vaynerchuk

16,605 次观看 • 10 个月前

On a previous #TeawithGaryVee I was asked "What are your predictions business wise about Hollywood, the film industry, and AI" If you have a question for me, drop it here so my team can add it to the next Rapid Fire Q&A and follow me on

On a previous #TeawithGaryVee I was asked "What are your predictions business wise about Hollywood, the film industry, and AI" If you have a question for me, drop it here so my team can add it to the next Rapid Fire Q&A and follow me on

Gary Vaynerchuk

14,280 次观看 • 1 年前

On a previous #TeawithGaryVee I was asked "How would you handle being the main supporter for your family, but you hate what you do?" if you have a question for me, drop it here so my team can add it to the next Rapid Fire Q&A and follow me on

On a previous #TeawithGaryVee I was asked "How would you handle being the main supporter for your family, but you hate what you do?" if you have a question for me, drop it here so my team can add it to the next Rapid Fire Q&A and follow me on

Gary Vaynerchuk

22,596 次观看 • 3 个月前

Here's a video of our Dell Technologies 7975 Precision workstation running local AI for coding agents. It features dual RTX6000 Blackwell GPU w/ 96GB each, so I'm running Qwen 3B on it. I had ChatGPT's desktop app set it all up and configure the integration into VS Code as well. I've done that by hand once before, and once was enough! Now I can query my local agent just as easily as I can with Codex or any other paid model. It's a bit slower, and I only use it to REVIEW code, not author it. Still, it's impressive that you can do this 100% locally now! I'm giving it a final pass over the Task Manager TMOG codebase on "Very High", with unlimited runtime! Only the best for :-)

Here's a video of our Dell Technologies 7975 Precision workstation running local AI for coding agents. It features dual RTX6000 Blackwell GPU w/ 96GB each, so I'm running Qwen 3B on it. I had ChatGPT's desktop app set it all up and configure the integration into VS Code as well. I've done that by hand once before, and once was enough! Now I can query my local agent just as easily as I can with Codex or any other paid model. It's a bit slower, and I only use it to REVIEW code, not author it. Still, it's impressive that you can do this 100% locally now! I'm giving it a final pass over the Task Manager TMOG codebase on "Very High", with unlimited runtime! Only the best for :-)

Dave W Plummer

25,646 次观看 • 16 天前