We have added support for OpenAI's GPT-Realtime-2 model in ChatBotKit. The model can now be selected and assigned to a bot in the same way as other supported language models, making it easier to experiment with realtime-capable AI while keeping the familiar bot configuration workflow.
This update allows GPT-Realtime-2 to operate through the normal bot completion flow, including text responses and function calling support. It also lays the groundwork for richer realtime experiences in the future without requiring a separate bot or integration type.
For teams building conversational workflows, this means realtime-capable models can now fit into the same model assignment and deployment patterns already used across ChatBotKit.