InterLock proxy that governs what AI agents can read and write in databases, files and APIs, with redaction and approvals.
DimOS
Python SDK and agent-native operating system for programming humanoid, quadruped and drone robots without ROS.
What is DimOS?
DimOS, from Dimensional, is a robotics software stack for building physical applications entirely in Python on humanoids, quadrupeds and drones. It installs with a single script, does not require ROS, and organizes a robot's software into modules that exchange standard messages over streams. Ready-to-run bundles called blueprints cover navigation and mapping, perception, spatial memory and agentic control, so you start a stack with one dimos command. You can replay recorded sessions with no hardware, run MuJoCo simulation, or connect a real Unitree Go2 over its IP address. LLM agents run as native modules and robot skills are exposed over MCP, so you can tell a robot in plain language to explore a room. A hosted teleoperation service, dimTELE, lets you drive robots from a browser or Quest headset. The documentation shows no pricing, and the project is described as open source.
Use cases
- Program a quadruped robot in Python
- Command a robot in natural language
- Map and navigate a room with SLAM
- Simulate a humanoid in MuJoCo
- Replay recorded robot sessions
- Teleoperate a robot from a browser
- Expose robot skills to coding agents over MCP
Who it's for
Why use DimOS
- You can write robot applications in plain Python instead of learning a ROS setup.
- Replay and simulation modes let you develop and test without owning a robot.
- Natural-language agents and MCP tools make it easy to connect coding agents to robot skills.
- One CLI covers starting, monitoring, logging and stopping a stack.
- dimTELE needs no inbound ports on the robot network, so remote driving works behind a home router.
DimOS features
- Python SDK with no ROS required
- Navigation with SLAM, obstacle avoidance and route planning
- Perception with detectors, VLMs and audio processing
- Spatio-temporal spatial memory
- LLM agents that run as native modules
- MCP server exposing robot skills
- dimos CLI for running, inspecting and stopping stacks
- MuJoCo simulation and recorded-session replay
- dimTELE hosted remote teleoperation over WebRTC
- Browser cockpit with camera, map and keyboard control
DimOS integrations
How to get started with DimOS
- Run the official install script, which sets up system packages, Python 3.12 and dimOS.
- Start a blueprint with dimos run, using the replay flag first so no hardware is needed.
- Switch to MuJoCo simulation, or set ROBOT_IP and run the blueprint against a real robot.
- Add an LLM agent and send commands such as exploring the room, or call MCP tools directly.
DimOS limitations
- Hardware support varies: Go2 is marked stable while G1, xArm and Piper are beta and others alpha or experimental
- The default agent uses OpenAI gpt-4o, so it needs an OPENAI_API_KEY unless you configure a local model
- Quickstart needs a machine with Ubuntu 22.04+ or macOS 14+, roughly 10 GB free disk and 16 GB RAM
- A GPU is needed for perception and AI features
DimOS pricing
DimOS is open source, so you can use and self-host it for free.
Prices can change and may vary by country or exclude tax. Check the official pricing page.
DimOS alternatives
The main DimOS alternatives listed on Debuly are Stagehand, Browser Use, Herdr, Letta, Moli and ZenML. The table compares pricing and platforms; order is A–Z.
| Tool | Pricing | Starting price | Platforms |
|---|---|---|---|
| DimOS | Open source | — | CLI, API |
| Browser Use | Paid | $0.02, usage-based | API, Web, CLI |
| Herdr | Open source | — | macOS, Linux, Windows, CLI |
| Letta | Freemium | $20/mo | CLI, Web, API |
| Moli | Open source | $12 one-time | Web, API, Python, Node.js |
| Stagehand | Open source | — | API |
| ZenML | Open source | $39/mo | Self-hosted, Cloud (SaaS), Docker, Python |
Browser Use
Cloud browsers and hosted web agents for AI, plus an MIT-licensed open-source browser agent, billed pay-as-you-go from $0.02 per hour.
Herdr
Open-source terminal runtime that keeps many coding agents running in the background and lets you reattach from any machine.
Open sourceVisit Herdr (official site)
Letta
Platform, formerly MemGPT, for building stateful AI agents with persistent memory, from a terminal app, SDK or API.
FreemiumVisit Letta (official site)
Moli
Open-source browser engine from Lexmount that lets AI agents extract web content and automate pages with low memory use.
Open sourceVisit Moli (official site)
Stagehand
Open-source SDK from Browserbase for building browser agents with act, extract and observe calls in TypeScript, Python or Go.
Open sourceVisit Stagehand (official site)
ZenML
Open-source orchestration for Python ML pipelines and AI agents, with Kitaru for recording and replaying agent runs against production data.
Open sourceVisit ZenML (official site)
DimOS FAQ
How much does DimOS cost?
No price is listed in the documentation. The software is installed from a public script and repository, and no paid tiers are described for the SDK itself.
Is DimOS open source?
Yes, it is described as open source, with its code in a public GitHub repository and installed from a script. Read the repository license before commercial use.
Does DimOS require ROS?
No. The docs say it works with a simple install and no ROS required, although navigation can also run through ROS.
Which robots does DimOS support?
The docs list the Unitree Go2 as stable, the Unitree G1, xArm and AgileX Piper as beta, MAVLink and DJI Mavic as alpha, and Unitree B1 and force torque sensors as experimental.
Does DimOS have an API or MCP support?
Yes. Every robot skill is exposed over MCP, and the dimos CLI can list and call tools, for example dimos mcp list-tools. Coding agents such as Claude Code can use those interfaces.
Can I try DimOS without a robot?
Yes. Replaying a recorded Unitree Go2 session needs no hardware or GPU, and simulation runs in MuJoCo.
What are alternatives to DimOS?
Alternatives include ROS 2, NVIDIA Isaac, Viam, Foxglove and Unitree's own SDKs, depending on whether you need a full robotics framework or only tooling. On Debuly, compare Stagehand, Browser Use and Herdr.
Sources & verification
Facts on this page come from the official sources below, each with the date we read it. Written by Debuly; the vendor did not review this page.
| Source | Type | Accessed |
|---|---|---|
| dimensional.org | official | |
| github.com/dimensionalOS/dimos | repository | |
| docs.dimensional.org/ | docs | |
| docs.dimensional.org/quickstart/ | docs | |
| docs.dimensional.org/usage/cli/ | docs |
Image credits
- Logo: from framerusercontent.com · fetched · Vendor trademark, used only to identify the product; removed on request
- Cover: none collected yet, so a generated tile is shown.
Spot an error? Report an issue with this listing. We also remove a logo or image on the owner's request.
More in Developer platforms 4
All developer platforms toolsAI-native feature flag platform that wraps risky pull requests in flags, ramps rollouts on error signals and opens cleanup PRs.
FreemiumOpen source notebook that writes many candidate prompts for a task, tests them on your cases and ranks them with ELO scores.
Open sourcePython library for steering language models, combining constrained generation with ordinary control flow so outputs follow regexes, grammars or JSON schemas.
Open source
