MacPaw partners with Liquid AI for on-device Mac inference

The interesting part of the MacPaw / Liquid AI announcement is not the partnership headline, it is where it puts the inference. According to TechCrunch, the Ukraine-based Mac developer is building a locally hosted version of its Eney assistant on top of an on-device inference system Liquid AI is calling Elix, paired with a local memory system.

The pitch from Liquid AI's co-founder and CEO Ramin Hasani is that the model architecture is chosen for the hardware first. 'Before training our models, we select an architecture that is different and tailored to the hardware,' he told TechCrunch, arguing that lets 'the most efficient version of intelligence' run directly on the device. MacPaw CEO Oleksandr Kosovan frames the user-side payoff plainly, saying locally hosted models will let people 'run assistants and agentic workflows offline.'

The larger move is Setapp. MacPaw's subscription app store has over 150,000 paying users, and the plan is to eventually open the same on-device inference stack to developers building for that store, so their Mac apps can call a local model instead of paying per-token to a cloud API. TechCrunch reports MacPaw will experiment with credit-based pricing, where users spend credits per AI operation based on task complexity, and that once the local architecture is finalized developers will also get a path to cloud models from providers like Google.

The honest caveat is that the reporting is thin on the parts that matter most for anyone trying to plan around this. There is no ship date for the locally hosted Eney, no detail on how credits are split between MacPaw and third-party developers, and no benchmark for what Liquid AI's on-device models can actually do relative to hosted frontier systems. If it lands as described, the winners are Mac indies who want to add AI features without underwriting an OpenAI bill, and the interesting long game is whether MacPaw can define the Mac's local-inference layer before Apple decides to define it for everyone.