DEV Community

Muhammad Mairaj
Muhammad Mairaj

Posted on

How to build a Claude Sonnet 4.5 agent in 10 minutes

We are going to build an AI agent that uses Claude Sonnet 4.5. We will use the agent primitive by Langbase.

An agent works as a runtime LLM agent. You can specify all parameters at runtime and get the response from the agent. The agent uses Langbase’s unified LLM API to provide a consistent interface for interacting with 600+ LLMs across all the top providers.

See the full list of supported models.

Okay, let’s get to building.

Step 1: Install Langbase SDK

Langbase SDK comes in both TypeScript and Python.

For TypeScript:

npm install langbase
Enter fullscreen mode Exit fullscreen mode

For Python:

pip install langbase
Enter fullscreen mode Exit fullscreen mode

Step 2: Get your Langbase API Key

Every request you send to Langbase needs an API key. To generate one:

  • Sign up at Langbase.com
  • From the sidebar, click API Keys
  • Create a new API key

For more details, follow the API key guide.
.

In your .env file, add:

LANGBASE_API_KEY=your_langbase_api_key
Enter fullscreen mode Exit fullscreen mode

Since we’re using Claude Sonnet 4.5 (Anthropic), you’ll also need your Anthropic API key:

LLM_API_KEY=your_anthropic_api_key
Enter fullscreen mode Exit fullscreen mode

Step 3: Write the Code

TypeScript version (index.ts):

const { output } = await langbase.agent.run({
    model: 'anthropic:claude-sonnet-4-5',
    instructions: 'You are a helpful AI Agent.',
    input: 'Explain what an AI Engineer does.',
    apiKey: process.env.LLM_API_KEY!,
    stream: false,
});

console.log(output);
Enter fullscreen mode Exit fullscreen mode

Run it:

npx tsx index.ts
Enter fullscreen mode Exit fullscreen mode

Python version (index.py):

from langbase import Langbase
import os
from dotenv import load_dotenv

load_dotenv()
langbase = Langbase(api_key=os.getenv("LANGBASE_API_KEY"))

response = langbase.agent.run(
    model="anthropic:claude-sonnet-4-5",
    instructions="You are a helpful AI Agent.",
    input="Explain what an AI Engineer does.",
    api_key=os.getenv("LLM_API_KEY"),
    stream=False,
)

print("response:", response.get("output"))
Enter fullscreen mode Exit fullscreen mode

Run it:

python index.py
Enter fullscreen mode Exit fullscreen mode

Step 4: That’s it 🚀

That’s basically 5–6 lines of code.

With Langbase’s agent primitive, you can:

  • Dynamically modify the prompt, model, or instructions
  • Enable streaming responses
  • Switch between 600+ AI models with the same API
  • Use both open-source (DeepSeek, Kimi K2, Grok 4) and frontier models (Claude Sonnet 4.5, GPT-5, Gemini 2.0, etc.)

What are you waiting for? Build your Claude Sonnet 4.5 agent!

Top comments (0)