Open Interpreter is an open-source local AI agent and terminal AI agent that lets large language models run code directly on your machine.¹⁾ As a fully open source AI agent with over 63,000 stars on GitHub, it provides a ChatGPT-like terminal experience that can execute Python, JavaScript, Shell, and more — with full access to your local filesystem, internet, and installed packages.

Unlike cloud-based code interpreters with sandboxed environments, Open Interpreter removes all restrictions on runtime, file size, and package availability, giving LLMs direct access to your computer's full capabilities including data analysis, browser control, media editing, and GUI interaction.²⁾

Open Interpreter splits into two core components: a Core execution engine and a Terminal Interface. The Core provides a real-time code execution environment where LLMs safely control the computer via an exec() function that takes a language identifier and code string. The Terminal Interface connects to LLMs via LiteLLM, streaming model messages, code blocks, and system outputs as Markdown.³⁾

The system supports any LLM provider through LiteLLM — including OpenAI, Anthropic, local models via LM Studio, and dozens more. Conversations can be persisted, restored, and run asynchronously.

• Local execution — No sandbox restrictions, full internet access, any package
• Multi-language support — Python, JavaScript, Bash, and more
• GUI and vision control — Interact with desktop GUI and analyze images
• LLM flexibility — Any provider via LiteLLM (OpenAI, Anthropic, local models)
• Chat persistence — Save and restore conversation sessions
• AGPL-3.0 licensed — Fully open-source with 100+ contributors⁴⁾

# Install Open Interpreter
# pip install open-interpreter
 
# Python API usage
from interpreter import interpreter
 
# Basic conversation
interpreter.chat("What operating system are we on?")
 
# Configure model
interpreter.model = "gpt-4o"
 
# Use with Anthropic
interpreter.model = "claude-3-5-sonnet-20240620"
interpreter.chat("Analyze the CSV files in my Downloads folder")
 
# Local/offline mode with LM Studio
interpreter.offline = True
interpreter.llm.model = "openai/x"
interpreter.llm.api_base = "http://localhost:1234/v1"
interpreter.llm.context_window = 3000
interpreter.llm.max_tokens = 1000
interpreter.chat()

%%{init: {'theme': 'dark'}}%%
graph TB
    User([User]) -->|Natural Language| TI[Terminal Interface]
    TI -->|Markdown Stream| User
    TI -->|Messages| LM[LiteLLM Router]
    LM -->|API Calls| OpenAI[OpenAI API]
    LM -->|API Calls| Anthropic[Anthropic API]
    LM -->|API Calls| Local[Local Models]
    TI -->|Code Request| Core[Core Engine]
    Core -->|exec| Python[Python Runtime]
    Core -->|exec| JS[JavaScript Runtime]
    Core -->|exec| Shell[Shell / Bash]
    Core -->|Results| TI
    Core -->|File Access| FS[Local Filesystem]
    Core -->|Network| Internet[Internet Access]

• Data analysis — Process CSV, Excel, databases with natural language queries
• File management — Organize, rename, convert files across directories
• Web automation — Control browsers, scrape data, interact with APIs
• Media editing — Resize images, convert formats, edit video metadata
• System administration — Monitor processes, manage services, configure systems⁵⁾

• Goose — AI Coding Agent by Block
• Continue.dev — Open-Source AI Code Assistant
• Chainlit — Conversational AI Framework