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.

With iOS 27 Apple introduced its most powerful on-device model: AFM 3 Core Advanced It’s a 20 billion parameters model activating only between 1 to 4 billion parameters depending on the request Here’s the model running on iPhone 17 Pro

Adrien Grondin

8,501 subscribers

472,311 views • 1 month ago •via X (Twitter)

Education Health & Wellness Science & Technology

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

Here’s Apple’s foundation model running on iPhone 17 Pro It’s just so fast Apple was not joking, the A19 Pro chip is really good for running LLMs on-device

Here’s Apple’s foundation model running on iPhone 17 Pro It’s just so fast Apple was not joking, the A19 Pro chip is really good for running LLMs on-device

Adrien Grondin

1,566,094 views • 1 year ago

How does Edge0 run a 35B model with only 1–2.5 GB of peak memory? In simple terms: • The full model stays in storage. • Only the parameters needed for the current step are loaded into memory. • The next parameters are prepared while the current step is running. • Memory is reused instead of loading the entire model at once. Large model. Small working set. Fully on-device.

How does Edge0 run a 35B model with only 1–2.5 GB of peak memory? In simple terms: • The full model stays in storage. • Only the parameters needed for the current step are loaded into memory. • The next parameters are prepared while the current step is running. • Memory is reused instead of loading the entire model at once. Large model. Small working set. Fully on-device.

Samuel Zeng

34,026 views • 1 month ago

Another demo of Apple 20B on-device foundation model that comes with iOS 27 The new models support image understanding (on all iPhone with the new OS) Very good at OCR

Another demo of Apple 20B on-device foundation model that comes with iOS 27 The new models support image understanding (on all iPhone with the new OS) Very good at OCR

Adrien Grondin

58,553 views • 1 month ago

> be Muse Spark Open > Meta's Llama 4 disappoints everyone > Meta goes closed with Muse Spark > Zuck writes a $100-200 billion check > Muse Wonder ships on Monday > 3 days later Meta drops you with open weights > 816 billion parameters > only 42 billion active at a time > 1 million token context > built for coding and long agent tasks > runs on two Mac Studios on a desk > Meta calls you the most capable US open-weight model ever > every Chinese lab reading the model card America is back in the open-weight game.

> be Muse Spark Open > Meta's Llama 4 disappoints everyone > Meta goes closed with Muse Spark > Zuck writes a $100-200 billion check > Muse Wonder ships on Monday > 3 days later Meta drops you with open weights > 816 billion parameters > only 42 billion active at a time > 1 million token context > built for coding and long agent tasks > runs on two Mac Studios on a desk > Meta calls you the most capable US open-weight model ever > every Chinese lab reading the model card America is back in the open-weight game.

Kirill

129,325 views • 21 days ago

Apple wasn’t kidding, the iPhone 17 Pro is really built for running LLMs Here’s LFM2 8B A1B by running on-device with MLX in Locally AI - Local AI Chat, the iPhone runs the 8B model with zero struggle Thanks Prince Canuma for the port to MLX, it made the MLX Swift port possible

Apple wasn’t kidding, the iPhone 17 Pro is really built for running LLMs Here’s LFM2 8B A1B by running on-device with MLX in Locally AI - Local AI Chat, the iPhone runs the 8B model with zero struggle Thanks Prince Canuma for the port to MLX, it made the MLX Swift port possible

Adrien Grondin

493,561 views • 1 year ago

I wanted to try Maple-Preview by DeepGrove on iPhone so I simply ported the model to MLX Swift using Bionic Maple runs at ~80tk/s on iPhone 17 Pro with my quick port, it’s a 20B-A1B model at 2-bit The model is close to Qwen3.5 35-A2B in many benchs and runs easily on iPhone

I wanted to try Maple-Preview by DeepGrove on iPhone so I simply ported the model to MLX Swift using Bionic Maple runs at ~80tk/s on iPhone 17 Pro with my quick port, it’s a 20B-A1B model at 2-bit The model is close to Qwen3.5 35-A2B in many benchs and runs easily on iPhone

Adrien Grondin

13,376 views • 2 months ago

The new PrismML’s Bonsai 27B model running on iPhone 17 Pro Bonsai 27B is the first 27B-class model to run on a phone and it uses less than 5GB of memory Pushing the limits on what can run a phone

The new PrismML’s Bonsai 27B model running on iPhone 17 Pro Bonsai 27B is the first 27B-class model to run on a phone and it uses less than 5GB of memory Pushing the limits on what can run a phone

Adrien Grondin

67,095 views • 3 months ago

Happy Friday! We just put DeepSeek-V4-Pro up on It’s the world’s largest open source model at 1.6T parameters, and you can run it for free running on NVIDIA Blackwell GPUs. Try the NVIDIA NIM API →

Happy Friday! We just put DeepSeek-V4-Pro up on It’s the world’s largest open source model at 1.6T parameters, and you can run it for free running on NVIDIA Blackwell GPUs. Try the NVIDIA NIM API →

NVIDIA AI

208,311 views • 6 months ago

Guys, I translated the model into English, but unfortunately, for some reason, the parameters were hidden when I added it - I'm so sorry! Here’s how to translate the model into English and reveal the customization parameters.

Guys, I translated the model into English, but unfortunately, for some reason, the parameters were hidden when I added it - I'm so sorry! Here’s how to translate the model into English and reveal the customization parameters.

D.maf

31,151 views • 25 days ago

Side by side comparison of Apple’s on-device foundation model iPhone 16 Pro vs 17 Pro ~55 tk/s for the 16 Pro and ~85 tk/s for the 17 Pro

Side by side comparison of Apple’s on-device foundation model iPhone 16 Pro vs 17 Pro ~55 tk/s for the 16 Pro and ~85 tk/s for the 17 Pro

Adrien Grondin

907,443 views • 1 year ago

These guys just released a 3.6B model that matches Qwen3-8B on formal logic with less than half the parameters.

These guys just released a 3.6B model that matches Qwen3-8B on formal logic with less than half the parameters.

me

131,149 views • 1 month ago

Google’s Gemma 4 E2B running on-device on iPhone 17 Pro Gemma 4 is built from the same research as Gemini 3, has image understanding capabilities and can reason if needed Running at ~40tk/s with MLX optimized for Apple Silicon

Google’s Gemma 4 E2B running on-device on iPhone 17 Pro Gemma 4 is built from the same research as Gemini 3, has image understanding capabilities and can reason if needed Running at ~40tk/s with MLX optimized for Apple Silicon

Adrien Grondin

1,046,050 views • 7 months ago

1/ Any model on OpenRouter can now run code in a hosted Linux container. Add one tool to a Responses API request. The model writes the script, runs it, and reads back the result. Nothing to install, nothing to host. Meet openrouter:shell. To get started: "tools": [{ "type": "openrouter:shell", "parameters": { "engine": "openrouter" } }]

1/ Any model on OpenRouter can now run code in a hosted Linux container. Add one tool to a Responses API request. The model writes the script, runs it, and reads back the result. Nothing to install, nothing to host. Meet openrouter:shell. To get started: "tools": [{ "type": "openrouter:shell", "parameters": { "engine": "openrouter" } }]

OpenRouter

39,516 views • 1 month ago

Grok 4 remains the world's most advanced and powerful AI model. 😎

Grok 4 remains the world's most advanced and powerful AI model. 😎

SMX 🇺🇸

9,891,150 views • 1 year ago

🚨 do you understand what happened to Apple.. Apple just announced Siri AI at WWDC 2026. The new "revolutionary" assistant is literally Google Gemini running in a costume. - Apple pays Google 1 billion dollars a year for a custom 1.2 trillion parameter Gemini model - the only Apple part of Siri AI is the logo and the privacy marketing - Siri AI launches in fall 2026 and only works on iPhone 15 Pro and newer - Tim Cook leaves as CEO on September 1, three weeks before it ships the company that used to be the most valuable on earth couldn't build its own AI and had to rent one from the search engine.

BuBBliK

144,839 views • 5 months ago

The iPhone 18 Pro is much better than what I expected for on-device AI It can run a 27B model at double the speed of the 17 Pro The new A20 Pro chip is a beast

The iPhone 18 Pro is much better than what I expected for on-device AI It can run a 27B model at double the speed of the 17 Pro The new A20 Pro chip is a beast

Adrien Grondin

427,294 views • 1 month ago

The new Laguna S 2.1 model by Poolside is now free for 2 weeks on Nous Portal. At 118B total parameters with 8B active, it's quick to run and the most capable model they've released so far. Try Portal today at

The new Laguna S 2.1 model by Poolside is now free for 2 weeks on Nous Portal. At 118B total parameters with 8B active, it's quick to run and the most capable model they've released so far. Try Portal today at

Nous Research

196,298 views • 3 months ago

DeepSeek V4 Flash Running 12 GB phone CPU only. - Model: 284B parameters - Speed: 1 tok/s - Just clever expert streaming from flash storage. -

DeepSeek V4 Flash Running 12 GB phone CPU only. - Model: 284B parameters - Speed: 1 tok/s - Just clever expert streaming from flash storage. -

Md Ismail Šojal 🕷️

28,301 views • 2 months ago