I’m a little reluctant to pay money for access to AI capabilities while the marketplace seems to be going through some billing fluctuations, with speculation about charges going up as providers realise how dependent companies have become.
For the time being, I’m okay with trying out running some LLM capabilities on my own hardware. I’ve started out with Gemma 4 from Google DeepMind, which I have set up to run inside Docker.
https://deepmind.google/models/gemma/gemma-4
I installed it a few weeks ago, but recently got around to playing with it a bit more by pointing an IDE (Xcode) at it for use for its AI chat capability.
Today I also decided to have a go at developing some software that interacts with the LLM as a capability included within a program. So far this just involves connecting to the model, sending it a chat prompt and then outputting back the response, but as I get more familiar with langchain4j I expect to make something much more sophisticated.
Leave a Reply