The storage statistics are pretty cool in the latest #Damus update @Damus

😎

One bit and ternary AI image generation on my phone. It actually works quite well. It is wild they can losslessly compress these large models down enough to work well on a phone.

What will Prism do next - Audio, video, translation, transcription?

#ai

https://prismml.com/news/bonsai-image-4b

They also released a ternary model that only slightly increases the size while greatly improving accuracy.

#ai #llm

https://prismml.com/news/ternary-bonsai

Microsoft has been looking at something similar #ai #llm

https://arstechnica.com/ai/2025/04/microsoft-researchers-create-super%E2%80%91efficient-ai-that-uses-up-to-96-less-energy/

Been playing around with this model locally on my phone. So far I’m impressed with how well it works. #ai #llm

8 billion parameters in just 1.2 GB of space. For comparison Qwen 3.5 takes up just over 3 GB for 4 billion parameters.

It will be interesting to see where these new 1bit models lead.

https://www.prnewswire.com/news-releases/prismml-launches-worlds-first-1-bit-ai-model-to-redefine-intelligence-at-the-edge-302730568.html

RIP Chuck Norris