When Token market is shown in your portal (/inference), use it as described below. If you do not see it, it is not open for your account yet. When the inference service is open, chat completion requests can use stream: true. Responses arrive as server-sent events, allowing your client to display text as it is generated. Token market (/inference) provides streaming examples for curl and the OpenAI Python SDK. Streaming uses the same billing approach as a standard response. A configured generation time limit still applies; streaming does not remove request limits or make an unavailable model available.
Does inference support streaming?
From the EverXCloud help center.
