Liquid Apollo

Liquid AI

About this app

Apollo is a platform for discovering and running new AI language models right on your phone. Explore, rate, and use language models that run right in your hand, with no internet connection at all. Powered by the Liquid Edge AI Platform (LEAP).

Use Apollo to:
- Draft messages, emails, with AI even when offline.
- Chat with a secure, private AI without the information leaving your device.
- Craft interesting digital characters.
- Send images to LFM2-VL models to interact with the visual world.

Ratings and reviews

I love the Liquid models! For their size they have a great quality. They're fast too. The app is a little buggy though, kinda not the best for navigation as well, and the context length could be better especially for smaller models that can support it. I'm not really sure whether it's using the search etc. or not, there's no feedback/info. Lastly, it would be nice to be able to set the download location, and even use other manual local models if possible (maybe not).

Came here after watching the PBS episode. Got two suggestions: - It seems like web search does not work. It would be great if you could fix that. The October 2023 cutoff could be mitigated a bit by this I guess. - Persistent memory would have been interesting, i.e. AI model learning about the user and through the earlier interactions, maybe even continue on earlier chats.

this service is awesome, the only caveats are that sometimes it tells me that my device doesn't have Google play, when it clearly does and recently it deleted all of my chat history without warning and without any way to recover it. I believe in the vision of on device AI so much, and this is the most straight forward way about it. PS. for some reason, LFM 2.5 1.2B doesn't load on LM studio.

What’s new

• New frosted-glass toolbar with edge-to-edge chats
• Redesigned Settings with instant Dark/Light/System switching
• Pause, resume, and delete downloads with one tap
• Fixed models disappearing from "All models" after downloading one version
• Chat header now shows the exact model version you're using