October 7, 2024

Release date: October 7, 2024

Overview

In Release 71, we've introduced these new features:

• Advanced methods for knowledge model chat API
• UX improvements
• Support for new large language models

What's new

Advanced methods for knowledge model chat API

We’ve expanded the knowledge model chat API with new chat methods to enhance interactions with AI models. These methods include the following:

• Creating new chat sessions.
• Sending messages and retrieving session lists.
• Accessing message history with metadata like chat modes, message counts, and document references.
• Specifying knowledge model IDs, choosing interaction modes (direct, blended, or RAG-only), and selecting specific LLMs.

This API enhancement provides developers with precise control over AI conversations, enabling the creation of dynamic and context-aware applications. With detailed session and message management, including metadata such as chat modes and document references, the API supports sophisticated, responsive, and trackable AI experiences. These improvements result in more effective knowledge retrieval and enhanced user engagement across a wide range of AI-integrated applications.

UX Improvements

Answers generated by Model now display the name and type of the model used. Additionally, users can provide feedback on the model's answers, enhancing usability and interaction quality.

Support for new large language models

Users can now select the following models for a specific chat thread in the knowledge model from the list of available models:

• o1-preview (GPT-o1 Preview) — a beta version of the o1 model, designed to perform complex reasoning and think before providing an answer.
• o1-mini (GPT-o1 Mini) — a lighter version of the o1 model, offering twice the context window size and faster response times.