Language model selection

Language models function as the assistant’s engine, converting instructions and knowledge into structured output. They’re trained on enormous amounts of text data to recognize patterns, reason and generate text in a way that resembles human language. Intric is a model-agnostic platform that supports both closed and open models, giving users the freedom to choose between different language models.

Factors for model selection:

1. Sensitivity: Are you handling sensitive information? Choose a model that’s operated in Sweden or the EU.
2. Purpose: Do you need speed or complex analysis (reasoning)? Test different models.
3. Security class: If a model doesn’t appear in the list, it’s likely because the Space’s security class doesn’t allow it (e.g., American models in a “Sensitive” Space).