Skip to content

Open Interpreter

AI Agents Open Interpreter 0.4.2+sushi.2 Hardened AMI

Open Interpreter is an Ubuntu 24.04 AMI that lets a language model write and run code for you from plain-English requests: analyze and chart data, clean up files, convert formats and automate scripts. It runs on an isolated instance in your own account instead of your laptop, asks before it runs any code, and works with the model provider you choose.

Get started View on AWS Marketplace
AMI release
0.1.6 (Open Interpreter 0.4.2+sushi.2)
Recommended
t3.medium
Storage
16 GiB+ encrypted gp3
Access
Session Manager or SSH
Released
Oct 12, 2026
Delivery
Amazon Machine Image in your own AWS account

Describe a task in plain English. The model writes Python or shell code, explains it and runs it on the instance to deliver the result.

Safe defaults

Every code run asks for your confirmation and auto-run is off. No provider key or listening port is included.

Isolated cloud instance with no listening ports and no baked keys, so your laptop stays out of reach. Use the model provider you choose.

Quick start

Full guide →
  1. Subscribe
  2. Launch the instance
  3. Connect
  4. Run the walkthrough
  5. Set your provider key
  6. Run your first task

What is included

Component Details
Operating system Ubuntu 24.04 LTS, x86_64
Open Interpreter 0.4.2, installed as the Software Sushi build 0.4.2+sushi.2 in its own Python 3.12 environment at /opt/open-interpreter/venv
Command /usr/local/bin/interpreter
Profile ~/.config/open-interpreter/profiles/default.yaml for the ubuntu user, created on first boot with auto-run off and telemetry disabled
Workspace /var/lib/open-interpreter/workspace, owned by ubuntu
Self-checks A walkthrough at /opt/sushi/docs/demo.sh and a self-test at /opt/sushi/bin/smoke.sh. Neither needs a model key.

The +sushi.2 build keeps Open Interpreter's Python code unchanged. It updates the pinned versions of some dependencies in the package metadata. The source and license notices are under /usr/share/doc/open-interpreter.

There is no web UI and no background service. No model, model download or provider key is included: you bring your own.

Safe defaults

  • Confirmation first. The profile sets auto_run: false, so Open Interpreter shows each code block and runs it only after you approve.
  • No -y by accident. The interpreter command refuses -y and --auto_run unless you allow it for a session. See Allow auto-run for one session.
  • Telemetry off. The profile sets disable_telemetry: true.

The upstream command still checks PyPI for newer versions when it starts. When you exit, it asks whether you want to send feedback; nothing is shared unless you answer yes.

Ports

Open Interpreter does not listen on any port. You do not need any inbound security group rule. The instance needs outbound HTTPS for Session Manager and for your model provider.

  • t3.medium (2 vCPU, 4 GiB RAM).
  • At least 16 GiB encrypted gp3 root volume.
  • IMDSv2.

Billing

You pay an hourly software fee for each running instance, shown on the AWS Marketplace listing. An optional five-day free trial covers the software fee for one instance and then converts to hourly billing. It is the same software before and after the trial. EC2, EBS, data transfer and your model provider's charges are billed separately, including during the trial.

Where to go next

Upstream project

Open Interpreter is published under the AGPL-3.0 license. Software Sushi packages it independently and does not claim upstream endorsement. The corresponding source is on the instance under /usr/share/doc/open-interpreter.

Last updated October 14, 2026 · Suggest an improvement