Last updated 07-22-2026
Category:
Reviews:
Join thousands of AI enthusiasts in the World of AI!
Open Interpreter
Open Interpreter is an open-source coding agent that runs in your terminal or as a desktop app. Open a project, type interpreter, and it reads your files, edits code, and runs commands with sandbox guardrails. It is built on a fork of OpenAI's Codex CLI and stays provider agnostic, so you can plug in OpenAI, Anthropic, Kimi, Qwen, DeepSeek, or local models.
The project targets low-cost and open-weight models rather than frontier-only agents. Model-specific harnesses emulate the tool interfaces each provider trained on, which the team reports improves benchmark scores on models like Kimi K2.5 and DeepSeek V4 Flash compared to generic agent loops.
Developers get a familiar Codex-style terminal with AGENTS.md support, skills, MCP servers, and session management. The separate desktop app brings the same agent workflow to Word, Excel, PDFs, and the browser for knowledge workers who do not live in a code editor.
Model-specific harnesses tuned for Kimi, Qwen, DeepSeek, and other open models
OS-level sandbox on macOS, Linux, and Windows with approval gates before network access
Swap providers and models from the terminal with
/modelwithout reinstalling anythingKeeps Codex skills, MCP servers, AGENTS.md, and slash commands from the upstream fork
Desktop app fills PDF forms, edits Word docs, and builds Excel sheets with built-in editors
Apache 2.0 open source with a mature Codex-derived terminal interface.
Supports dozens of model providers plus local Ollama and LM Studio setups.
Native OS sandboxing reduces the need for Docker-only isolation during autonomous runs.
Desktop tier starts free if you bring your own API keys or sign in with ChatGPT.
Harness emulation is reported to improve open-model benchmark scores versus generic agent loops.
Hosted desktop models route requests through Open Interpreter servers with up to 30-day logging.
Managed model access on desktop requires Pro ($20/mo) or Business ($60/mo) subscriptions.
The terminal product is CLI-first, which may feel steep for users who avoid the command line.
Enterprise plan details beyond usage multiples require contacting sales directly.
What is Open Interpreter?
Open Interpreter is a coding agent that runs in your terminal. You open a project, type `interpreter`, and it can read files, edit code, run commands, and ask for approval before escalating beyond the current sandbox.
Which models can I use with Open Interpreter?
Open Interpreter supports built-in providers like OpenAI, Amazon Bedrock, Ollama, and LM Studio, plus hosted providers such as Anthropic, OpenRouter, DeepSeek, Moonshot/Kimi, Z.ai, Alibaba/Qwen, and Groq. You can also connect custom providers with OpenAI-compatible or Anthropic Messages-compatible APIs.
How does the Open Interpreter sandbox work?
Open Interpreter uses OS-level sandboxing on macOS, Linux, and Windows. Permission profiles control read access, workspace writes, and network use. When a command or tool needs approval, the terminal UI shows the request before it runs.
Can I run multiple Open Interpreter sessions at once?
Yes. Open Interpreter supports multiple parallel sessions. You can open one terminal tab per session and use `/resume`, `/fork`, or `/side` inside the TUI, or run `interpreter resume` and `interpreter fork` from the shell.
Is Open Interpreter open source?
Yes. Open Interpreter is open source under the Apache 2.0 License. The terminal product is a fork of Codex and keeps the Codex terminal surface while remaining provider agnostic.
Is my code private when using Open Interpreter?
Local models stay on your machine. If you use your own API key, requests go directly from your device to that provider. If you use hosted models through Open Interpreter, requests are routed through their servers and logged for up to 30 days per their privacy policy.
Does Open Interpreter have a desktop app?
Yes. Open Interpreter offers a desktop app for Word, Excel, PDFs, browsers, and local files. The desktop tier has Free, Pro ($20/mo), and Business ($60/mo) plans listed on the product page.