Google's new open source LLM that is supposedly very good for coding and math tasks. I'm running it on my MacBook and it's incredibly quick due to it's relatively small size (there is a 2B and a 7B model available).
So far I found Google's (free-tier) Gemini to make more mistakes during coding than (paid) GPT. Rumors are that Google's paid Gemini model is incredibly large and good and has a context window size of 10M tokens (40 times of GPT 4?) but this is just what I've heard, I've never had access to it.
Nevertheless, it's always good to start with running your own LLM, and this seems like an accessible model to do so because of, you know, this little thing called dystopia.
https://ollama.com/library/gemma
Just starting out on local models myself. Found Claude and GPT to be far more consistent than Gemini. Giving autogen studio a whirl this week.
I am sticking with codeLlama or Mistral. Not trusting Google...!
free 2mo trial of gemini pro is out
way more accurate on complex medical questions, but then ragequit 🤣🤦♂️