We use multiple versions of Llama2 and Llama3 which are used as needed by the system. Simple model for simple needs.

LLama 3 was also used to power conversations on hillda by choice of a user

Their open-source model really ignited the use of this product and is one of the most capable open-source models.

This is the second incredible AI for creating texts. For some types of texts it is even better than GPT-4, which is why we use it.

Open source Llama 3, combined with Groq is just amazingly fast and performant.