> ## Documentation Index
> Fetch the complete documentation index at: https://docs.vectrade.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Streaming

> Real-time streaming with Server-Sent Events (SSE) for AI analysis and live data.

# Streaming

VecTrade uses Server-Sent Events (SSE) for real-time streaming of AI analysis responses.

<Warning>
  **SDK AI Streaming is coming soon.** Direct `client.ai.stream()` is not yet available in the public SDK. AI streaming is currently available through the [VTrade Copilot](/guides/vtrade/copilot) web interface for Standard and Professional plan users.
</Warning>

## AI Streaming

<CodeGroup>
  ```python Python theme={null}
  from vectrade import VecTrade

  client = VecTrade()

  # Stream AI analysis
  for chunk in client.ai.stream("What's the bull case for NVDA?"):
      print(chunk.text, end="", flush=True)

  print()  # newline after stream ends
  ```

  ```typescript TypeScript theme={null}
  import { VecTrade } from '@vectrade/sdk';

  const client = new VecTrade();

  const stream = client.ai.stream('What\'s the bull case for NVDA?');
  for await (const chunk of stream) {
    process.stdout.write(chunk.content);
  }
  ```

  ```bash cURL theme={null}
  curl -N -H "Authorization: Bearer $VECTRADE_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{"prompt": "Bull case for NVDA?", "stream": true}' \
    https://api.vectrade.io/v1/ai/analyze
  ```
</CodeGroup>

## SSE Event Format

The stream sends events in standard SSE format:

```
data: {"content": "NVIDIA's", "type": "text"}

data: {"content": " bull case", "type": "text"}

data: {"content": " rests on", "type": "text"}

data: {"type": "citation", "source": "Q4 Earnings", "url": "..."}

data: {"type": "done", "usage": {"prompt_tokens": 150, "completion_tokens": 420}}

```

## Stream Event Types

| Type       | Description                      |
| ---------- | -------------------------------- |
| `text`     | Content chunk to display         |
| `citation` | Source reference                 |
| `metadata` | Model info, timing               |
| `done`     | Stream complete with usage stats |

## Async Streaming (Python)

```python theme={null}
import asyncio
from vectrade import AsyncVecTrade

async def main():
    client = AsyncVecTrade()

    async for chunk in client.ai.stream("Analyze TSLA technicals"):
        print(chunk.text, end="", flush=True)

    await client.close()

asyncio.run(main())
```

## Edge/Worker Streaming (TypeScript)

For Vercel Edge Functions or Cloudflare Workers, convert to a ReadableStream:

```typescript theme={null}
import { VecTrade, toReadableStream } from '@vectrade/sdk';

export const runtime = 'edge';

export async function POST(req: Request) {
  const { prompt } = await req.json();
  const client = new VecTrade();

  const stream = client.ai.stream(prompt);
  const readable = toReadableStream(stream);

  return new Response(readable, {
    headers: { 'Content-Type': 'text/event-stream' },
  });
}
```

## Error Handling in Streams

Errors during streaming are delivered as error events before the connection closes:

```python theme={null}
from vectrade import VecTrade, RateLimitError

try:
    for chunk in client.ai.stream("..."):
        print(chunk.text, end="")
except RateLimitError as e:
    print(f"\nRate limited: retry after {e.retry_after}s")
```

## Connection Management

* Streams auto-close when the `[DONE]` event is received
* Use context managers for guaranteed cleanup:

```python theme={null}
with VecTrade() as client:
    for chunk in client.ai.stream("..."):
        print(chunk.text, end="")
# Connection is closed automatically
```