OpenHands

OpenHands is for people who want Devin-style autonomy and the source. You hand it an issue, it works in its own environment and comes back with a pull request rather than a chat transcript, and the same agent is reachable from a terminal CLI, the Agent Canvas GUI, a Python SDK or the hosted cloud. The daily trait that decides it is that the output is a branch you review.

Expect a setup evening and a token bill on a real repository. Pick it if you want to self-host and audit every step. Pick Devin if you would rather someone else run the machine and send the invoice.

The worst thing is the meter, and it is your meter. A run that loops on a failing test bills every attempt, and billing at cost is not the same as a cap; nothing in the product stops a bad afternoon from costing what a good week would. Autonomy reads until it stops, and you pay for the reading.

The consequence is that a budget alarm belongs in your provider console before the first run, not after the first invoice. What it does right: commands execute in a Docker sandbox, so a bad rm is a container's problem and your working tree is one volume mount away from safe.

OpenHands is the one tool on this board whose headline claim can be checked. The 71.8 percent on SWE-bench Verified with GPT-5 is a public leaderboard submission dated 2025-08-07 and checked by the benchmark maintainers, so it is documented rather than self-reported, and comparable with every other entry that went through the same harness.

Two caveats. 1. A scaffold-plus-model result attributes the score to the pair; swap the model and the number is no longer yours. 2. Verified is the curated subset, so the figure says nothing about the harder residue. The observation: the scaffold is the reproducible half, and the half most vendors decline to publish.

OpenHands has the distribution: roughly 86,000 stars, an open-source license the field forks freely, and a benchmark entry rivals cite. The business model is where I squint. The Individual cloud plan is free for one user with ten conversations a day, which is a customer-acquisition subsidy paid in inference, and Enterprise is custom priced, so the revenue line is whatever a sales team can negotiate above zero.

Likely acquirer: a cloud provider that wants an agent platform with a community already attached, or a lab that wants the scaffold. Position: long the project, neutral the company, and keep a fork warm.

The demo is an agent closing an issue by itself. Enterprise offers self-hosted in our VPC, SAML SSO and bring-your-own-key, which answers the data-residency question in one line; the price is custom, so the answer costs a sales call. The free path is sixty engineers each running the stack locally with sixty personal keys and no central audit log, which is the configuration I would find on the laptops if I did not act first.

Onboarding is a day per engineer, mostly installation. Approved with conditions: Enterprise quote in hand, self-hosted, a spend limit per seat, and the free path blocked by policy.

OpenHands is the full stack and I own all of it. MIT license, the local-LLM docs cover any OpenAI-compatible endpoint so my vLLM box is a URL in a config, MCP servers go in config.toml under [mcp] as stdio_servers or shttp_servers, and the Python SDK lets me drive the agent from my own scripts instead of its UI. Every layer is a file I can read.

Forkability is real: the project has already survived a rename, and the code does not care what it is called. The cost is a setup weekend and a machine that can carry it. That is ownership, with assembly required.