How-to: Installing AI to L5 and running it locally offline with ollama.ai

I think that varies - completely open and free for any use is (as with a lot of software and services) a challenge and rarity. Some models have different licenses. Hopefully this will get better and some of the worst offenders get left behind (OpenAI and ChatGPT are very problemmatic). Regarding AI openness, one of the FOSDEM talks showed these slides, which add to your list in the AI context:

Here are a couple of links I found regarding lisences and AI. It pretty much boils down to that open models are usable but some may have limitations (usually the very large ones regarding commersial use). See: Top 10 List of Large Language Models in Open-Source | Deci and OpenRAIL: Towards open and responsible AI licensing frameworks and Licensing Machine Learning models — The Turing Way. There is also Dolly (Free Dolly: Introducing the World's First Truly Open Instruction-Tuned LLM), which allows also commercial use freely, but is not presently in ollama library.

[edit to add: ollama itself is MIT license]
[edit to add2: Microsoft’s research has also developed the phi model, which is small(ish) and has several derivatives like llama, and it uses the MIT license (see bottom) - free but from M$, pick your poison]
[edit to add3: noticed that also available as small(ish) version is StarCoder which has it’s own AI-relevant license]

2 Likes