OpenHands

Open-source autonomous AI software engineer that writes, runs, and debugs code

Hand-reviewed by the Toolglade team·Updated Aug 2026

coding#ai-coding-agent#open-source#autonomous-agent#swe-bench

Free plan Free trial Claimed API Self-hosted Teams

About OpenHands

OpenHands is an open-source autonomous coding agent that plans, edits, runs, and debugs code in a sandboxed runtime, available self-hosted for free or via a managed cloud, and is model-agnostic across major LLM providers.

OpenHands treats a coding agent as a full autonomous developer: given a task, the agent plans, edits files, executes shell commands, runs tests, and browses documentation inside a sandboxed runtime, iterating until the task is done. It is model-agnostic, so teams can point it at Claude, GPT, Gemini, or open-weight models, and it ships with a web UI, a CLI, and a headless API for integration into CI and automation. The project rebranded from OpenDevin to OpenHands in late 2024 under the All-Hands-AI GitHub organization and has since become one of the largest open-source AI coding communities, with a rapid contribution cadence and a strong showing on SWE-bench-style benchmarks. All Hands AI raised an $18.8M Series A to fund development. Beyond the free MIT-licensed self-hosted core, the company offers OpenHands Cloud (a hosted individual tier with BYOK or metered model usage) and an Enterprise edition that adds a control plane for scheduled and event-triggered automations, an internal add-on marketplace, role management, API keys, secrets storage, and cost visibility.

TL;DR

OpenHands is a leading open-source autonomous coding agent from All Hands AI, offering a free self-hosted core and a managed cloud, and is model-agnostic across major LLM providers.

Company overview

OpenHands is developed by All Hands AI, the company behind the open-source project formerly known as OpenDevin. After rebranding to OpenHands in late 2024, it built one of the largest open-source AI coding communities on GitHub and raised an $18.8M Series A to accelerate development.

The company monetizes through OpenHands Cloud, a hosted service, and an Enterprise edition with governance and automation features, while keeping the core framework free and MIT-licensed to sustain community adoption.

Product features

The product pairs an autonomous agent loop with a sandboxed runtime where the agent can edit files, run shell commands, execute tests, and browse the web. It exposes a web UI, CLI, and headless API, and is model-agnostic across Claude, GPT, Gemini, and open weights.

Enterprise capabilities add a control plane for scheduled and event-triggered automations, an internal add-on marketplace, role management, API keys, secrets storage, and cost visibility dashboards for teams running agents at scale.

Target market

OpenHands targets software developers, OSS maintainers, and engineering teams that want an open-source, model-flexible coding agent they can self-host or run in a managed cloud, plus enterprises needing governance around agent automation.

Buyer personas

End users

Software engineers and OSS maintainers running the agent to fix bugs, build features, and automate repo tasks.

Buyers

Engineering leaders and platform teams evaluating open-source agent tooling and enterprise governance.

Key influencers

Open-source contributors, AI researchers, and developer advocates in the agent community.

Ideal customer profile

Technically capable teams that value open-source, model flexibility, and self-hosting, and are comfortable managing LLM API costs.

Funding & performance

All Hands AI raised an $18.8M Series A. Verify the latest funding details with the vendor or public sources.

Pros & cons

Pros

Fully open-source (MIT) with a large, active community
Model-agnostic — works with Claude, GPT, Gemini, and open weights
Strong SWE-bench performance for an open project
Sandboxed runtime keeps agent actions isolated
Free self-hosted option plus a hosted cloud tier
Headless API for CI and automation

Cons

Autonomous runs can consume significant model tokens and cost
Self-hosting requires Docker and some setup effort
Agent reliability varies with task complexity and chosen model
Enterprise features are quote-only
Less polished than commercial IDE-integrated tools

Pricing plans

Open Source (Self-Hosted)

MIT-licensed core
Local sandboxed runtime
BYO model API keys
CLI and web UI
Headless API

Cloud (Individual)

$0 / month

Hosted OpenHands Cloud
BYOK or metered model usage at cost
GitHub/GitLab integration
No infra to manage

Enterprise

Custom

Control plane for automations
Role management and RBAC
Secrets storage and API keys
Cost visibility dashboard
Internal add-on marketplace

Key features

API

Team collaboration

Self-hosted

Multi-language

Integrations

GitHub, GitLab, Anthropic, OpenAI, Google Gemini, Docker

Input types

text

Output types

text

Best For

autonomous coding tasks, bug fixing, open-source AI dev workflows

Compare key features

View all alternatives →

Feature

OpenHands

Refact.ai

LiteLLM

Pricing

Freemium

Free plan

Yes

Free trial

Yes

API

Yes

Self-hosted

Yes

Team support

Yes

Frequently asked questions

Is OpenHands free?+

Yes. The self-hosted version is free and open-source under the MIT license, and there is a free individual cloud tier. You pay only for the underlying model usage unless you self-host with your own keys.

What was OpenHands called before?+

It was originally released as OpenDevin and rebranded to OpenHands in late 2024 under the All-Hands-AI organization.

Which LLMs does OpenHands support?+

It is model-agnostic and works with Anthropic Claude, OpenAI GPT, Google Gemini, and open-weight models via LiteLLM-style configuration.

Can OpenHands run autonomously?+

Yes. It can plan and execute multi-step tasks — editing files, running commands, and testing — inside a sandboxed runtime, with human review of the resulting diffs.

Does OpenHands integrate with GitHub?+

Yes. It can connect to GitHub and GitLab repositories to resolve issues and open pull requests, and the Enterprise edition adds scheduled and event-triggered automations.