At a glance
- Category
- Coding agents
- Tags
- Open source, Self-hosted
Code
- Last push
- 15 October 2026
What is OpenHands?
OpenHands is an open-source, model-agnostic platform for building and running autonomous AI coding agents. Formerly known as OpenDevin, it provides a Software Agent SDK, CLI, web GUI, and hosted cloud service for software engineering tasks like writing code, fixing bugs, reviewing pull requests, and migrating legacy codebases. With 71,000+ GitHub stars, MIT licensing, and support for dozens of LLM providers, OpenHands works as the leading self-hostable alternative to proprietary coding agents like Devin or GitHub Copilot.
Agent Canvas is a local workspace that hosts other agents
The Agent Canvas page calls it "the new face of OpenHands": "a local, visual workspace for running AI coding agents", installed on macOS, Linux or Windows. It runs "a single agent-server on your laptop", with "Docker is supported, but optional", and parallel agents each run "isolated in its own git worktree".
Canvas reaches agent harnesses through the open Agent Client Protocol. The page marks four as supported: Claude Code, Codex and Gemini CLI on "your existing subscription", and the OpenHands Agent with "bring your own LLM key". The home page repeats the claim, "Use Claude Code, Codex, Gemini CLI, OpenHands, or your own stack", and says it runs open weights such as Kimi, DeepSeek and Qwen "locally, in the cloud, or inside your VPC". Canvas also "ships with a library of MCP connections, Agent Skills", and the MCP docs describe MCP as the way the agent talks to external tool servers. The agent uses MCP as a client. OpenHands documents no MCP server of its own.
Three plans on the price table, and every one is self-hosted
The pricing page now names its free column Open Source rather than Agent Canvas: "Open source (MIT)", "Runs on your laptop or shared VM", "Single user / shared identity". Teams reads "$500/month" with "10 users included (+$50/mo per seat)" and a 30-day trial, one organization, multi-user identity and RBAC, SSO / SAML, budgeting controls and built-in integrations. Enterprise is "Custom pricing" with unlimited seats and users, an enterprise agreement and a customer success manager.
The comparison table's Deployment row reads "Local Self-hosted" for Open Source and "Self-hosted" for Teams and Enterprise. A note below it says "The Teams plan is capped at 5,000 sessions per month. Enterprise usage remains uncapped." Built-in OAuth integrations on Open Source are "Select MCP only". Support runs from "Open source community" through "Best effort email" on Teams to "SLA-based response times" on Enterprise. Every column carries "Bring your own LLM key", so model spend is a separate bill that the page does not estimate. Only monthly figures are shown, with no annual billing option.
A managed cloud appears on the site but not on the price table
OpenHands Cloud is offered without a price. The Agent Canvas page lists it as a third backend, "Always-on cloud agents in isolated sandboxes" with "One-click integrations" and "Team sharing", labeled "Optional · managed cloud", and says you can "switch backends from the header without touching your config". The home page promises Team users can "Keep agents running when laptops are closed".
The pricing page names no cloud plan and marks Teams and Enterprise as self-hosted. The Cloud getting-started docs call it "the hosted cloud version of OpenHands", reached at app.all-hands.dev by logging in with GitHub, GitLab or Bitbucket, and name no plan, cap or price. The Cloud API docs describe a REST API for starting and checking conversations with a bearer key.
Automations open pull requests, and review stays with engineers
Automations are the main pitch beyond chat. The home page lists templates triggered from GitHub, Slack, Jira, Linear and schedules: reviewing labeled pull requests, triaging Linear issues and Jira incidents, drafting incident retrospectives from Notion and Jira, fixing CI failures and remediating security findings, each ending in a comment or a pull request.
The site describes the human role in two ways. The Enterprise page heads its automation section "From event to PR, no human in the loop." and two sentences later says "Automation doesn't mean unsupervised", with every run "fully logged and governed by the same cost and security policies as manual sessions". The home page says agents take the repetitive work "while keeping engineers in control of review, approval, and merge". Agents open the pull request, and the vendor's pages leave the merge to a person. The Enterprise control plane lists role-based access and SSO, policy enforcement, centralized audit logs and budget management.
The security answer depends on where OpenHands runs
Its public controls include "Customer data deleted upon leaving" and "Data retention procedures established".
The home page FAQ says OpenHands "runs in your environment, whether on-prem or private cloud, so your code never leaves your control". That describes the self-hosted deployments. The hosted cloud runs on the vendor's own infrastructure, with the subprocessors listed above.
Where it fits, and where it does not
The Enterprise page addresses engineering organizations that want to run Claude Code, Codex or open models "on AWS, Azure, or Google Cloud while keeping code, data, and execution within your security boundaries", with one control plane for access, audit and spend. A solo developer gets the MIT Open Source plan free and can keep an existing Claude Code, Codex or Gemini CLI subscription.
It fits less well when a buyer wants one predictable bill or a hosted service with a price. Every plan needs a separate model provider account, and the managed cloud has no published price. On Teams, each user past ten costs $50 a month, and usage above 5,000 sessions a month means moving to custom Enterprise pricing. Teams support is best effort email, and SLA response times are reserved for Enterprise.
Key Features
- Software Agent SDK: A composable Python library for defining and running agents locally or at scale, with model-agnostic orchestration through the litellm library
- Sandboxed Execution: All code runs inside secure Docker containers, so agents can write, test, and iterate on code without risking your local environment
- Model-Agnostic Support: Works with OpenAI, Anthropic Claude, Google Gemini, local models through Ollama or vLLM, and 75+ other providers through litellm
- GitHub and GitLab Integration: Agents clone repositories, create branches, push commits, and open pull requests directly within existing repo workflows
- Local GUI and CLI: A browser-based React interface with embedded VS Code editor, terminal access, and a separate CLI for terminal-first workflows
- OpenHands Cloud: Hosted deployment at app.all-hands.dev with Slack, Jira, and Linear integrations, plus a free tier with 10 daily conversations
- Skills System: Specialized prompts with domain-specific knowledge that let teams adapt agents to particular workflows without rebuilding core logic
- SWE-Bench Performance: Resolves 53%+ of real-world GitHub issues on SWE-bench Verified when paired with strong models, scoring 77.6 on the benchmark
Use Cases
- Engineering teams handling maintenance backlogs: Use OpenHands to automate repetitive tasks like fixing merge conflicts, resolving linter failures, refactoring modules, and triaging GitHub issues across multiple repositories
- Backend developers migrating legacy codebases: Run agents to navigate large codebases, convert code between languages (COBOL to Java, for example), and test changes across multiple modules with limited manual work
- DevOps teams running automated security workflows: Set up agents for vulnerability scanning that automatically generate pull requests with fixes, integrated into existing CI/CD pipelines
- Solo developers prototyping applications: Use the Planning Agent to scope features from a high-level prompt, review a generated plan, then let the agent handle implementation, testing, and debugging
- Not ideal for: Non-technical users who need point-and-click automation (use Make, Zapier, or n8n instead) or teams that need pre-tuned domain-specific agents with strict compliance requirements
Strengths and Weaknesses
Strengths:
- Fully open-source with MIT licensing, so teams can self-host, audit, and modify without vendor lock-in
- Model-agnostic design means you can swap between cloud providers or run local models based on cost and privacy needs
- Active development with 71,000+ stars, 490+ contributors, and monthly releases including recent additions like Plan Mode and slash menus
- Sandboxed Docker execution keeps agent actions isolated and auditable, a real advantage for enterprise security requirements
- The free open-source tier has no conversation limits, and the cloud individual tier starts at $0 with pay-as-you-go LLM usage at cost
Weaknesses:
- Not fully autonomous yet, as users report needing to provide corrections and guidance during longer tasks
- Sandbox setup can be tricky, with reports of crashes and outdated documentation after the v0.10 release
- Rapid release cadence has introduced breaking changes in agent protocols, which can frustrate teams on older versions
- Error messages are sometimes vague, including reports of tasks failing without useful tracebacks
- The free cloud tier is limited to 1 workspace and 10 daily conversations
FAQ
Is OpenHands free?
Yes. The open-source version is completely free with no conversation limits. The cloud individual tier is also free with 10 daily conversations and pay-as-you-go LLM costs at provider rates with no markup.
What was OpenHands previously called?
OpenHands was previously called OpenDevin. The project rebranded and has since grown to 71,000+ GitHub stars.
Does OpenHands support local models?
Yes. OpenHands works with local models through Ollama, vLLM, and any provider supported by litellm. You can run it entirely on your own infrastructure with no external API calls.
How does OpenHands compare to Cursor?
OpenHands handles autonomous, long-running software engineering tasks in a sandboxed environment with support for 75+ LLM providers. Cursor is better for IDE-based coding with deep codebase indexing and real-time suggestions. Choose OpenHands for agent-driven automation; choose Cursor for daily editing.
Can OpenHands run code safely?
Yes. All code execution happens inside secure Docker containers. The sandbox isolates agent actions from your host system, and enterprise deployments can run in private VPCs with full RBAC controls.
How do I install OpenHands?
Clone the GitHub repo, run make build and make dev with Docker. On first launch, configure your LLM API key and model. Most developers report getting a working setup in 15 to 30 minutes.
OpenHands price and terms
Terms
- Free plan
- YesYes
The terms in full
Agent Canvas is free and MIT-licensed, self-hosted on your laptop or a shared VM, single user or shared identity, bring your own LLM key; no RBAC, SSO or budgeting. The paid Teams plan also offers a 30-day trial.
For AI agents
- MCP server
- Yesdocs.openhands.dev,
- API
- Yesdocs.openhands.dev,

Pricing page, captured
openhands.dev/pricing, checked
All plans and limits
| Plan | Price | Main limits |
|---|---|---|
| Open Source | $0 a month | Users: Single user / shared identity |
| Teams | $500 a month | 5,000 sessions per month; 10 users included |
| Enterprise | Price on request | Sessions: uncapped; Unlimited seats |
Read from openhands.dev/pricing.
What we checked on OpenHands
What the price counts
Flat monthly rate for Teams: $500 a month covers 10 users, extra seats cost $50 a month each, Enterprise is custom quoted, and the open-source edition is free.
$500/month 10 users included (+$50/mo per seat)
openhands.dev/pricingChecked
Free to use
Two standing free plans with no sales call: the local open-source build, and a free hosted Individual tier on the cloud.
Bring your own key or use our providers at-cost
openhands.dev/pricingChecked
Public API
A documented REST API starts and polls conversations with a bearer key, and the pricing table ticks Cloud APIs on all three plans.
OpenHands Cloud provides a REST API that allows you to programmatically interact with OpenHands.
MCP server
MCP is fully supported on CLI, SDK, local GUI and Cloud, letting the agent consume external tool servers over SSE, SHTTP or stdio.
Model Context Protocol (MCP) is an open standard that allows OpenHands to communicate with external tool servers, extending the agent's capabilities…
Open source
MIT licence in its public repository
MIT License
github.com/OpenHands/OpenHands/blob/mai…Checked by us,
Questions about OpenHands
- How much does OpenHands cost?
- OpenHands's Open Source plan is free. Its paid plans are Teams at $500 a month and Enterprise. From openhands.dev/pricing, checked 18 November 2026.
- Does OpenHands have an API or an MCP server?
- OpenHands has an API. From docs.openhands.dev/openhands/usage/cloud/cloud-api.md, checked 21 September 2026. OpenHands has an MCP server. From docs.openhands.dev/overview/model-context-protocol.md, checked 21 September 2026.
- Is OpenHands open source, and can you host it yourself?
- OpenHands is open source. We checked github.com/OpenHands/OpenHands/blob/main/LICENSE ourselves on 15 November 2026. You can host OpenHands yourself.
- What have we checked on OpenHands?
- 5 facts, under these headings: What it costs to start, how you build with it.
- How do we check OpenHands?
- We hold 5 facts on OpenHands, quoted from 6 pages. Facts last checked 17 November 2026. Price last checked 18 November 2026. Last change found 15 November 2026.
Sources and changes
Last changed
Sources6
What changed1
- What it does
Repository
- Licence
- MIT
- Stars
- 90,997
- Last push
- 15 October 2026
OpenHands price history
Teams has cost $500 a month since 2 November 2026, on all 5 readings. Before that it was $500 a month.
| Price | Plan | First seen | Last seen | Readings | Source |
|---|---|---|---|---|---|
| $500 a month | Teams | 2 November 2026 | 20 November 2026 | 5 | Archived copy, 2 November 2026 |
| $500 a month | 25 November 2025 | 10 February 2026 | 2 | Archived copy, 25 November 2025 |
From 7 readings of OpenHands's pricing page: 3 Internet Archive copies from 25 November 2025 and 4 of our own checks since 2 November 2026. A date is when we read the page, not when the price changed.
From our research
- AI agent pricing benchmark: what MCP-ready tools chargeWhat 18 MCP-ready AI tools charge to start, read on each vendor's pricing page in November 2026: per second, per minute, per token, per seat or per month.
- AI agent tools that let you take your data outClineFlow, AgentMail and Automater let you take your data out in full, while UI Discovery, AgentQL and Arcmira do so only in part.
- AI support agents: what each costs per resolved chatPublished prices for 21 AI support agents, converted to one comparable number: cost per conversation the AI closed on its own. Verified 13 Aug 2026.
Is OpenHands yours?
Free, and the page then carries the verified owner mark. You confirm it with an email at openhands.dev. Claim this listing
Report an error if a fact is wrong.