comparison
Local alternative to Claude Code
OpenJet is a local-first terminal coding agent. Same workflow as Claude Code — read files, edit code, run approved shell commands, resume sessions — but the model runs on your machine and your code, prompts, and tool output stay there too.
OpenJet vs Claude Code
| Category | OpenJet | Claude Code |
|---|---|---|
| Model runtime | Configures a local model runtime on your machine and points the agent at it. | Hosted by Anthropic, with AWS Bedrock and Google Vertex AI as enterprise routes. |
| Install | One-line installer that profiles your hardware and sets the runtime up for you. | npm install or native installer, plus account login and a network connection. |
| Command execution | Shell commands are approval-gated by default. | Permission model with read-only defaults and approval for shell commands. |
| Session resume | Saved chat state and resume checkpoints under .openjet/state/. | Conversation compaction and resume support via the hosted service. |
| Where data lives | Prompts, repo content, tool output, and transcripts stay on your machine. | Prompts and tool output flow through the hosted API. |
Why pick OpenJet over a hosted coding agent
- Local model, local tools. Read, search, edit, and write files plus approval-gated shell, all in one terminal workflow against a local model.
- Setup is the product. OpenJet handles the local runtime, model path, and agent harness so you don’t spend day one wiring it together.
- Built for resumable work. Session manifests, transcripts, and checkpoints live on disk so an interrupted session picks back up.
- No API key, no per-seat bill. Once installed, the agent runs against your local runtime. No hosted dependency, no metered tokens.
Read the docs or see the enterprise page for team use.