A: Hello iangeles,

Thank you for your question. Firstly, let's clarify something important: no information in Straico is used to train models. We ensure that every user's information is protected by a security layer, and models are always accessed anonymously to prevent any traceability regarding user queries.

Regarding your list of questions:

1. Currently, you can establish communication with a model, and each model has a maximum context of words it supports. For example, more advanced models like Claude 3 Opus support up to 150K words, while Gemini Pro supports 98K words. We offer more than 20 models, each with special abilities tailored for specific purposes. For instance, Mistral is an exceptional model designed for role-playing scenarios, simulating conversations with experts in various fields such as law or psychology. To interact with the models, you can provide context or input to guide the model's behavior in response to your questions.

2. In fact, our API allows for a single interaction: you send a prompt and receive an answer.

I understand that your questions are leaning towards the functionality of agents. Our initial step was to introduce model interaction using templates and context content, and our next step involves deploying agents in the third quarter of the year.