About Weights and Watts

Priya Raman is the pen name of this site's editorial team. We use one name so the site has one consistent voice. There's no single person called Priya behind it, and we won't claim otherwise or invent a home lab we don't have.

What we cover

Two things that keep colliding. First, the free open-weight models people run themselves: Qwen, Gemma, Llama, Mistral, DeepSeek, gpt-oss and the rest, plus the tools that run them, such as Ollama, LM Studio and llama.cpp. Second, the big hosted frontier models, explained in plain terms so you can tell when a free local model is enough and when it isn't.

Readers are in the US, the UK, Canada and Australia, so prices start in US dollars and we note local prices where the seller lists them.

How we research

Model cards, official release notes, documentation and the download pages themselves. If Ollama's library says a model is 9.0 GB, that's the number you get, with a link and the date we checked. Memory estimates are arithmetic on those published figures, and we show the assumptions so you can redo the sums for your own machine. Where we haven't been able to confirm something, we say so.

How we approach it

• Numbers first. Gigabytes, context length, exact commands. Then the opinion.
• Your hardware, not ours. We write for the GPU or laptop you already own before suggesting anything new.
• Licenses matter. "Open" doesn't always mean you can use it for anything. We point to the license.
• No hype. A new release gets a plain answer to one question: does it change what you should run?

How the site is paid for

Free to read. We may show ads, and some links may become affiliate links that pay us a commission at no cost to you. That never changes what we recommend. See our affiliate disclosure.

Corrections

Models and tools update weekly. If something here is out of date, tell us with the source and we'll fix it and note the change.