Nebius integration added

docs.json

@@ -157,6 +157,7 @@
                   "server/services/llm/google-vertex",
                   "server/services/llm/grok",
                   "server/services/llm/groq",
+                  "server/services/llm/nebius",
                   "server/services/llm/nim",
                   "server/services/llm/ollama",
                   "server/services/llm/openai",

server/services/llm/nebius.mdx

@@ -0,0 +1,333 @@
+---
+title: "Nebius"
+description: "LLM service implementation using Nebius AI Studio's API"
+---
+
+## Overview
+
+`NebiusLLMService` provides access to Nebius AI Studio's language models, supporting streaming responses, function calling, and context management with high-performance inference capabilities and flexible model selection.
+
+<CardGroup cols={2}>
+  <Card
+    title="Nebius LLM API Reference"
+    icon="code"
+    href="https://reference-server.pipecat.ai/en/latest/api/pipecat.services.nebius.llm.html"
+  >
+    Pipecat's API methods for Nebius integration
+  </Card>
+  <Card
+    title="Example Implementation"
+    icon="play"
+    href="https://github.com/pipecat-ai/pipecat/blob/main/examples/foundational/14x-function-calling-nebius.py"
+  >
+    Complete example with function calling
+  </Card>
+  <Card
+    title="Nebius Documentation"
+    icon="book"
+    href="https://docs.studio.nebius.com/"
+  >
+    Official Nebius AI Studio documentation
+  </Card>
+  <Card
+    title="Nebius AI Studio"
+    icon="microphone"
+    href="https://studio.nebius.com/"
+  >
+    Access models and manage API keys
+  </Card>
+</CardGroup>
+
+## Installation
+
+To use Nebius services, install the required dependencies:
+
+```bash
+pip install "pipecat-ai[nebius]"
+```
+
+## Prerequisites
+
+### Nebius Account Setup
+
+Before using Nebius LLM services, you need:
+
+1. **Nebius Account**: Sign up at [Nebius AI Studio](https://studio.nebius.com/)
+2. **API Key**: Generate an API key from your studio dashboard
+3. **Model Selection**: Choose from available Nebius models (Qwen, Llama, etc.)
+4. **Resource Allocation**: Configure compute resources as needed
+
+### Required Environment Variables
+
+- `NEBIUS_API_KEY`: Your Nebius API key for authentication
+
+## Configuration
+
+### Basic Setup
+
+```python
+from pipecat.services.nebius import NebiusLLMService
+
+# Initialize the Nebius LLM service
+llm = NebiusLLMService(
+    api_key="your-nebius-api-key",
+    model="Qwen/Qwen3-30B-A3B-fast",
+    params={
+        "temperature": 0.7,
+        "max_tokens": 1000,
+        "top_p": 0.95,
+        "frequency_penalty": 0.0,
+        "presence_penalty": 0.0
+    }
+)
+```
+
+### Available Models
+
+Nebius AI Studio provides access to various language models:
+
+- **Qwen Models**: `Qwen/Qwen3-30B-A3B-fast`, `Qwen/Qwen3-70B-Instruct`
+- **Llama Models**: `meta-llama/Llama-3-8b-chat-hf`, `meta-llama/Llama-3-70b-chat-hf`
+- **Custom Models**: Contact Nebius for enterprise model deployment
+
+### Advanced Configuration
+
+```python
+from pipecat.services.nebius import NebiusLLMService
+
+llm = NebiusLLMService(
+    api_key="your-nebius-api-key",
+    base_url="https://api.studio.nebius.com/v1/",  # Custom endpoint if needed
+    model="Qwen/Qwen3-30B-A3B-fast",
+    params={
+        "temperature": 0.6,
+        "max_tokens": 2048,
+        "top_p": 0.9,
+        "frequency_penalty": 0.1,
+        "presence_penalty": 0.1,
+        "stream": True  # Enable streaming responses
+    }
+)
+```
+
+## Usage Examples
+
+### Basic Chat Completion
+
+```python
+import asyncio
+from pipecat.services.nebius import NebiusLLMService
+from pipecat.frames.frames import TextFrame
+
+async def main():
+    llm = NebiusLLMService(
+        api_key="your-nebius-api-key",
+        model="Qwen/Qwen3-30B-A3B-fast"
+    )
+
+    # Process a text frame
+    response = await llm.process_frame(
+        TextFrame("What are the key benefits of using Nebius AI Studio?")
+    )
+    print(response)
+
+asyncio.run(main())
+```
+
+### Function Calling
+
+```python
+from pipecat.services.nebius import NebiusLLMService
+
+# Define functions for the LLM to use
+functions = [
+    {
+        "name": "get_weather",
+        "description": "Get current weather information",
+        "parameters": {
+            "type": "object",
+            "properties": {
+                "location": {"type": "string", "description": "City name"}
+            },
+            "required": ["location"]
+        }
+    }
+]
+
+llm = NebiusLLMService(
+    api_key="your-nebius-api-key",
+    model="Qwen/Qwen3-30B-A3B-fast",
+    params={"functions": functions}
+)
+```
+
+### Streaming Responses
+
+```python
+async def stream_response():
+    llm = NebiusLLMService(
+        api_key="your-nebius-api-key",
+        model="Qwen/Qwen3-30B-A3B-fast",
+        params={"stream": True}
+    )
+
+    async for chunk in llm.run_llm("Tell me about AI applications"):
+        if chunk:
+            print(chunk, end="", flush=True)
+```
+
+## Pipeline Integration
+
+### Complete Voice Assistant Example
+
+```python
+import asyncio
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.services.nebius import NebiusLLMService
+from pipecat.services.deepgram import DeepgramSTTService
+from pipecat.services.cartesia import CartesiaTTSService
+from pipecat.transports.daily import DailyTransport
+
+async def run_bot():
+    # Initialize services
+    stt = DeepgramSTTService(api_key="your-deepgram-key")
+    llm = NebiusLLMService(
+        api_key="your-nebius-api-key",
+        model="Qwen/Qwen3-30B-A3B-fast"
+    )
+    tts = CartesiaTTSService(api_key="your-cartesia-key")
+    transport = DailyTransport(
+        room_url="your-daily-room-url",
+        token="your-daily-token"
+    )
+
+    # Create pipeline
+    pipeline = Pipeline([stt, llm, tts])
+
+    # Run the bot
+    runner = PipelineRunner()
+    await runner.run(pipeline, transport)
+
+asyncio.run(run_bot())
+```
+
+## Authentication
+
+### API Key Authentication
+
+```python
+# Set environment variable
+export NEBIUS_API_KEY="your-nebius-api-key"
+
+# Or pass directly in code
+llm = NebiusLLMService(api_key="your-nebius-api-key")
+```
+
+### Custom Headers
+
+```python
+llm = NebiusLLMService(
+    api_key="your-nebius-api-key",
+    model="Qwen/Qwen3-30B-A3B-fast",
+    headers={"Custom-Header": "value"}
+)
+```
+
+## Error Handling
+
+```python
+from pipecat.services.nebius import NebiusLLMService, NebiusException
+
+try:
+    llm = NebiusLLMService(
+        api_key="your-nebius-api-key",
+        model="Qwen/Qwen3-30B-A3B-fast"
+    )
+
+    response = await llm.process_frame(TextFrame("Hello"))
+
+except NebiusException as e:
+    print(f"Nebius API error: {e}")
+except Exception as e:
+    print(f"General error: {e}")
+```
+
+## Performance Optimization
+
+### Batching Requests
+
+```python
+# Enable request batching for better performance
+llm = NebiusLLMService(
+    api_key="your-nebius-api-key",
+    model="Qwen/Qwen3-30B-A3B-fast",
+    params={
+        "batch_size": 10,
+        "batch_timeout": 0.1
+    }
+)
+```
+
+### Connection Pooling
+
+```python
+# Configure connection pooling
+llm = NebiusLLMService(
+    api_key="your-nebius-api-key",
+    model="Qwen/Qwen3-30B-A3B-fast",
+    max_connections=20,
+    timeout=30.0
+)
+```
+
+## Troubleshooting
+
+### Common Issues
+
+1. **API Key Issues**
+   - Ensure your API key is valid and has sufficient credits
+   - Check that the API key has access to the selected model
+
+2. **Model Availability**
+   - Verify the model name is correct and available in your region
+   - Some models may require special access permissions
+
+3. **Rate Limiting**
+   - Implement exponential backoff for rate-limited requests
+   - Monitor your usage in the Nebius AI Studio dashboard
+
+4. **Streaming Issues**
+   - Ensure your network supports persistent connections
+   - Check firewall settings for WebSocket/SSE connections
+
+### Debug Mode
+
+```python
+import logging
+
+# Enable debug logging
+logging.basicConfig(level=logging.DEBUG)
+
+llm = NebiusLLMService(
+    api_key="your-nebius-api-key",
+    model="Qwen/Qwen3-30B-A3B-fast",
+    debug=True
+)
+```
+
+## Best Practices
+
+1. **Model Selection**: Choose the appropriate model size based on your latency and quality requirements
+2. **Parameter Tuning**: Adjust temperature and top_p values based on your use case
+3. **Error Handling**: Implement robust error handling for production applications
+4. **Monitoring**: Monitor API usage and response times in production
+5. **Security**: Never expose API keys in client-side code or public repositories
+
+## Support
+
+For additional support with Nebius integration:
+
+- [Nebius AI Studio Documentation](https://docs.studio.nebius.com/)
+- [Nebius Support Portal](https://support.nebius.com/)
+- [Pipecat Community Discord](https://discord.gg/pipecat)