Universal Clients

Agent SDK provides a unified interface for various LLM providers. You can switch between providers by simply changing the client class.

Supported Clients

Client Class	Provider	Description
OpenAIClient	OpenAI	Supports GPT-4o, GPT-3.5 Turbo, and compatible APIs (e.g., DeepSeek, Grok).
GeminiClient	Google	Supports Gemini 1.5 Pro, Flash, etc.
AnthropicClient	Anthropic	Supports Claude 3.5 Sonnet, Opus, Haiku.
OpenRouterClient	OpenRouter	Unified access to almost all open and closed models.
OllamaClient	Ollama	For running local models (Llama 3, Mistral, etc.).
DeepSeekClient	DeepSeek	Specialized client for DeepSeek V3 and R1.

Usage Examples

OpenAI

from agent_sdk import OpenAIClient

client = OpenAIClient(api_key="sk-...")

Google Gemini

from agent_sdk import GeminiClient

client = GeminiClient(api_key="AIza...")

Local Models (Ollama)

from agent_sdk import OllamaClient

# Default base_url is http://localhost:11434
client = OllamaClient(base_url="http://localhost:11434")

Direct Chat Usage (No Agent)

You can use clients directly without the Agent/Runner abstraction if you just need a simple chat completion.

Synchronous Chat

messages = [{"role": "user", "content": "Explain quantum physics."}]

# Non-streaming
response = client.chat(model="gpt-4o", messages=messages)
print(response["content"])

# Streaming
stream = client.chat_stream(model="gpt-4o", messages=messages)
for event in stream:
    print(event.data, end="")

Asynchronous Chat

import asyncio

async def main():
    stream = client.chat_stream_async(model="gpt-4o", messages=messages)
    async for event in stream:
        print(event.data, end="")

asyncio.run(main())