We are excited to announce that ChatBotKit now supports o3-deep-research and o4-mini-deep-research, OpenAI's purpose-built models for autonomous, multi-step research tasks. These models go beyond standard question answering by combining advanced reasoning with integrated, real-time web search - enabling AI agents to independently gather information, synthesize findings, and produce well-grounded research reports.

Unlike general-purpose language models, the deep research variants are specifically optimized to plan and execute extended research workflows. Given a complex question or topic, they can autonomously break the problem into subtasks, issue targeted web searches, evaluate sources, reconcile conflicting data, and deliver comprehensive answers - all within a single session.

Choosing the Right Deep Research Model

o3-deep-research is OpenAI's most capable deep research model, ideal for high-stakes or nuanced investigations where accuracy and depth matter most. It excels at synthesizing large volumes of information, navigating ambiguous or technical topics, and producing structured, citation-backed outputs suitable for professional and academic use.

o4-mini-deep-research offers the same core research capabilities at a significantly lower cost and faster response time, making it well-suited for teams running research-intensive workflows at scale or for applications that require frequent, iterative research queries. Both models always use web search as part of their execution, which contributes additional compute cost per request.

Both models are available now on ChatBotKit and can be selected when configuring bots, blueprints, or any AI workflow that benefits from grounded, real-world knowledge retrieval.