Anthropic

Build AI agents with Claude models and E2B sandboxes

This guide shows you how to connect Claude models to E2B sandboxes for code execution.

Latest Claude models

Model	API ID	Best for
Claude Sonnet 4.5	claude-sonnet-4-5-20250929	Complex agents and coding
Claude Haiku 4.5	claude-haiku-4-5-20251001	Fast, cost-efficient tasks
Claude Opus 4.5	claude-opus-4-5-20251101	Maximum intelligence

Simple code execution

Python

# pip install anthropic e2b-code-interpreter
from anthropic import Anthropic
from e2b_code_interpreter import Sandbox

client = Anthropic()

response = client.messages.create(
    model="claude-sonnet-4-5-20250929",
    max_tokens=1024,
    system="You are a helpful assistant that writes Python code. Only respond with the code to execute, no backticks or explanations.",
    messages=[{"role": "user", "content": "Calculate the first 10 Fibonacci numbers"}]
)

code = response.content[0].text

with Sandbox() as sandbox:
    execution = sandbox.run_code(code)
    print(execution.text)

Tool use (function calling)

Claude’s tool use lets the model decide when to execute code.

Python

# pip install anthropic e2b-code-interpreter
from anthropic import Anthropic
from e2b_code_interpreter import Sandbox

client = Anthropic()
model = "claude-sonnet-4-5-20250929"

tools = [{
    "name": "execute_python",
    "description": "Execute Python code in a Jupyter notebook cell and return the result",
    "input_schema": {
        "type": "object",
        "properties": {
            "code": {
                "type": "string",
                "description": "The Python code to execute"
            }
        },
        "required": ["code"]
    }
}]

messages = [{"role": "user", "content": "What's the sum of the first 100 prime numbers?"}]

# First API call - Claude decides to use the tool
response = client.messages.create(
    model=model,
    max_tokens=1024,
    tools=tools,
    messages=messages
)

messages.append({"role": "assistant", "content": response.content})

# Execute the tool if called
if response.stop_reason == "tool_use":
    tool_use = next(block for block in response.content if block.type == "tool_use")
    
    with Sandbox() as sandbox:
        execution = sandbox.run_code(tool_use.input["code"])
        result = execution.text

    messages.append({
        "role": "user",
        "content": [{
            "type": "tool_result",
            "tool_use_id": tool_use.id,
            "content": result
        }]
    })

    # Second API call - Claude provides the final answer
    final_response = client.messages.create(
        model=model,
        max_tokens=1024,
        tools=tools,
        messages=messages
    )
    print(final_response.content[0].text)

Extended thinking

Claude 4 models support extended thinking for complex reasoning tasks.

Python

from anthropic import Anthropic
from e2b_code_interpreter import Sandbox

client = Anthropic()

response = client.messages.create(
    model="claude-sonnet-4-5-20250929",
    max_tokens=16000,
    thinking={
        "type": "enabled",
        "budget_tokens": 10000
    },
    messages=[{
        "role": "user", 
        "content": "Write Python code to solve the traveling salesman problem for 5 cities using dynamic programming"
    }]
)

# Extract the code from the response (after thinking)
for block in response.content:
    if block.type == "text":
        code = block.text
        break

with Sandbox() as sandbox:
    execution = sandbox.run_code(code)
    print(execution.text)

Next steps

More LLM integrations

Connect other LLM providers to E2B

Upload files

Work with files in your sandbox