News for hackers. Stuff that matters.

darkolorin

darkolorin on Hacker News

Show HN: We made LM studio alternative based on own engine

Posted by darkolorin on Thursday October 23, 2025 @02:55PM
from the show-hn dept.

https://trymirai.com/chat-for-mac

Hi guys. We made own Mac app for M series hardware. Based on own engine called uzu (available with MIT on GitHub). Completely written from scratch and inference too. Our goal is to show what you can do with local models. In the app section we will publish use cases which you can easily implement in your app. Would appreciate any feedback.Right now there is no set of proper benchmarks from us....

( Read More... | no comments | 3 points | on HN )
Re:Show HN: We made our own inference engine for Apple Silicon on Tuesday July 15, 2025 @04:54PM (#44575711)
Basically “faster” means better performance e.g. tokens/s without loosing quality (benchmarks scores for models). So when we say faster we provide more tokens per second than llama cpp. That means we effectively utilize hardware API available (for example we wrote our own kernels) to perform better.
Re:90T/s on my iPhone llama3.2-1B-fp16 on Tuesday April 01, 2025 @10:46PM (#43553251)
I made it! 90 t/s on my iPhone with llama1b fp16

We completely rewrite the inference engine and did some tricks. This is a summarization with llama 3.2 1b float16. So most of the times we do much faster than MLX. lmk in comments if you wanna test the inference and I’ll post a link.

Re:Quantum product on Thursday November 27, 2014 @08:15PM (#8669306)
It's my first experience on Medium. I hope community can help me to improve my skills.
Re:Morning Mail – one email at a time on Thursday August 28, 2014 @07:54AM (#8236777)
really great idea, but why morning? %)