Free AI on Your Mac Just Got Faster — No Internet Needed

Ollama is a free piece of software that lets you download and run AI chatbots directly on your own computer, rather than typing questions into a website and sending your words off to a company's servers. It's popular with people who want AI help without paying a monthly subscription or handing over their private notes, emails and documents. This week the team behind it released version 0.40.0, an early test version of the next update.

The headline change is that on modern Macs — the ones with Apple's own M1, M2, M3 and M4 chips — Ollama now automatically uses something called MLX. Think of MLX as a special adapter Apple built so AI software can squeeze the most out of its own chips. Previously the app used a more generic approach that worked everywhere but wasn't tuned for Macs. The practical result: AI models that used to feel sluggish, like a video buffering on bad Wi-Fi, should now respond noticeably faster and smoother. If you have ever waited several seconds for a chatbot to finish a sentence, that wait shrinks.

For everyday users, the appeal is twofold. First, privacy: because the AI runs on your own machine, nothing you type is uploaded anywhere. That matters if you're summarising medical notes, contract details or personal journals. Second, cost: no per-message fees, no subscription. You just need a reasonably recent Mac with enough memory — the bigger, smarter models need more of it, and an older or low-RAM machine will struggle.

The catch is that this is a pre-release, meaning it's unfinished and may be buggy. It currently supports a limited set of AI models, with more being tested during this phase. It also only helps Mac users on newer Apple Silicon machines; Windows and older Intel Macs see no change here. Still, the direction is clear: capable, private AI that lives on your device is becoming the default, not a niche experiment.