A coding agent built as a new Rust version based on OpenAI's Codex, focused on optimizing agent harness performance for low-cost models. It runs commands inside native sandboxing on macOS, Linux, and Windows and requires external LLM provider credentials.
Project overview
The project emulates provider harnesses and supports multiple LLMs and protocols specifically to optimize agent performance for low-cost models.
Project type
AI Coding
Use cases
Coding & Development
Deployment
Refer to project documentation
License
Apache-2.0
Best for
Developers who require a coding agent optimized for low-cost models and who have external LLM provider credentials.
Key capabilities
Runs commands inside native sandboxing on macOS, Linux, and Windows.
Runs as an Agent Client Protocol (ACP) agent for compatible editors.
Keeps configuration and session state local under the user's home directory.
Limitations and risks
The original Python version is now maintained as a separate community fork, meaning the primary project is the Rust version.
The data boundary, cost dependencies, GPU requirements, and telemetry are not documented.
Getting started
The setup difficulty is rated easy, providing a one-line curl/PowerShell installation script as the first success path. No coding is required to install.
Evidence and sources
README: A coding agent optimized for low-cost models.
README: Open Interpreter is a fork of OpenAI's Codex, with a focus on emulating the agent harness that gets the best performance out of low-cost models.
README: - Runs commands inside native sandboxing on macOS, Linux, and Windows. - Switches providers and models from the TUI with `/model`. - Inspects or switches Rust-native model harness…
README: This is the new Rust version of Open Interpreter, based on Codex. Looking for the original Python project? It lives on as a community-maintained fork at [endolith/open-interpreter…
README: Open Interpreter works in [ACP-compatible editors and clients](https://agentclientprotocol.com/get-started/clients). Configure the client to launch `interpreter acp`; see the [ACP…