Streaming

Real-time response streaming.

Overview

Streaming allows responses to appear in real-time as they're generated, rather than waiting for the complete response.

How It Works

Model → Token → Token → Token → Complete
         ↓        ↓        ↓
      Display  Display  Display

Instead of waiting for the full response, tokens are sent as they're generated.

Configuration

{
  agent: {
    streaming: true,
    streamingChunkSize: 10  // Tokens per chunk
  }
}

Channel Support

Not all channels support streaming equally:

Channel	Streaming Support
WebChat	Full
Control UI	Full
Telegram	Partial (edit messages)
WhatsApp	Limited
Discord	Partial

Benefits

• Faster perceived response: Users see output immediately
• Better UX: More natural conversation feel
• Cancelable: Users can cancel mid-response

Limitations

• Some channels don't support message editing
• Network latency can affect smoothness
• Tool calls may pause streaming

See Also

• Agent Loop
• WebChat