Skip to content 
Client SDKs
Use the Python and TypeScript SDKs for the Letta API
The Letta API is a RESTful API at https://api.letta.com that provides programmatic access to stateful agents with persistent memory.
Letta provides official SDKs for Python and TypeScript that simplify API integration. All API endpoints are also available directly via REST.
Prerequisites
Section titled “Prerequisites”To use the Letta API, you’ll need an API key.
For a step-by-step V1 SDK tutorial, see the V1 SDK overview.
Authentication
Section titled “Authentication”All requests to the Letta API must include these headers:
| Header | Value | Required |
|---|---|---|
Authorization | Bearer <your-api-key> | Yes |
Content-Type | application/json | Yes |
If you are using the client SDKs, the SDK will send these headers automatically.
Why use the SDKs?
Section titled “Why use the SDKs?”We recommend developers building on the Letta API use the client SDKs, which provide:
- Automatic header management
- Type-safe request and response handling
- Built-in retry logic and error handling
Installation
Section titled “Installation”Python:
pip install letta-clientTypeScript:
npm install @letta-ai/letta-clientPython SDK
Section titled “Python SDK”from letta_client import Letta
client = Letta(api_key="your-api-key")
# Create an agentagent = client.agents.create( model="openai/gpt-4.1", embedding="openai/text-embedding-3-small",)
# Send a messageresponse = client.agents.messages.create( agent_id=agent.id, input="Hello!")TypeScript SDK
Section titled “TypeScript SDK”import Letta from "@letta-ai/letta-client";
const client = new Letta({ apiKey: "your-api-key" });
// Create an agentconst agent = await client.agents.create({ model: "openai/gpt-4.1", embedding: "openai/text-embedding-3-small",});
// Send a messageconst response = await client.agents.messages.create(agent.id, { input: "Hello!",});