@bryan.thebuilder: Day 12/14 • can local AI replace Gemini? Apple Intelligence could always do multimodal prompts (text + images) BUT only in Apple's own apps. iOS 27 is the first time that developers can access these APIs, so I decided to put it to the test. Although the results weren't as good as my current Gemini (+ Luna) stack, they were pretty promising for something that's free, offline, and runs purely on my two year old iPhone 16 Pro. It's crazy how quickly local AI is advancing. a few years ago this would have been a state-of-the-art model and now it's running on something that fits in my pocket. I still think this model has a lot of potential so I'm excited to see what people do with it in iOS 27!

Bryan Lin

Open In TikTok:

Region: US

Saturday 17 October 2026 16:10:49 GMT

3923

181

Music

Download

No Watermark .mp4 (5.72MB) Watermark .mp4 (5.33MB) Music .mp3

Comments

overtimefantasy :

How long does it take to process an image?

2026-10-21 19:01:18

Random Stuff :

Not a hotdog 🌭

2026-10-19 04:58:27

Arman Piric :

Dang 40%?! Do you think it requires additional change in how and what the task is structured to be more successful?

2026-10-19 13:13:43

muneeb :

good info bro! could be useful for shortcuts

2026-10-17 20:29:03

80s-Amul :

On device Ai? Bookmarked.

2026-10-19 13:38:02

vanzvrzn :

Good luck!

2026-10-18 23:04:58

To see more videos from user @bryan.thebuilder, please go to the Tikwm homepage.