Free Tool Now Lets Your Laptop AI 'Think Harder' on Demand

Ollama is free software that lets you run AI models directly on your own laptop or desktop, instead of paying a monthly fee or handing your data to a big company. Its newest version, 0.34.3, adds something surprisingly useful: a 'thinking' control. Some modern AI models can spend extra effort working through a problem before answering. Until now, you couldn't easily see or change that setting. Now you can.

Here's why that matters. Imagine asking an AI to plan a trip. On the 'low' setting, it blurts out an answer in seconds. On 'max,' it quietly reasons through options, checks its logic, and takes longer — but usually gets it right. This update lets you choose per model, either by typing one command or by checking a page on ollama.com. That's a real trade-off you now control: speed when you want it, quality when you need it.

The update also brings Nemotron H vision models — AI that can look at images — to Apple Silicon Macs, the chips inside recent MacBooks and Mac Minis. So Mac owners can now run picture-reading AI locally, without sending photos anywhere. There's also a small but welcome fix: Ollama's macOS app will no longer pop open windows you deliberately closed, and downloads from HuggingFace (a popular library of free AI models) should work more reliably.

The catch: this is a minor maintenance release, not a revolution. Most changes are aimed at people who already run AI locally — a technical, do-it-yourself crowd. If you only use ChatGPT or Claude in a browser, nothing changes for you today. And running powerful models on your own machine still needs a reasonably modern computer with enough memory.