Feature Overview

Below is a matrix of Deepgram's Voice Agent API features. Please refer to the corresponding documentation for more details.

Voice Selection

Supported LLM Models

Conversation Context

Inputs: Client Messages

Outputs: Server Events

Input / Output Events

Session Length

Sessions close automatically after 2 hours. The server warns you 5 minutes ahead with MAXIMUM_SESSION_LENGTH_APPROACHING and closes the session with a MAXIMUM_SESSION_LENGTH_REACHED error. To keep a conversation going, start a new session and replay the prior turns in agent.context. See Maximum session length.

Rate Limits

For information on Deepgram’s Concurrency Rate Limits, refer to our API Rate Limits Documentation.

Deepgram Self-Hosted

Having challenges with performance and latency? Check out Deepgram’s Self-Hosted Solution to get the benefits of running your own hosted instance of Deepgram.