Reason
A state-of-the-art coding agent for cost efficiency
847 followers
A state-of-the-art coding agent for cost efficiency
847 followers
Reason is a state-of-the-art coding for cost efficiency.
We scored higher than harnesses like Codex & Pi on benchmarks like Terminal Bench 4.0 and DeepSWE while being 22% cheaper.
Reason can work in the cloud or locally. Give it a task and it writes code, runs tests and opens a pull request. Use Reason from the Mac app, terminal, web or Slack, with your choice of models and connected subscriptions. Start free on your own computer or cloud!
This is the 6th launch from Reason. View more
Reason
Launching today
Reason is the best coding agent, in the cloud or locally. Give it a task and it writes code, runs tests and opens a pull request. Use Reason from the Mac app, terminal, web or Slack, with your choice of models and connected subscriptions. Start free on your own computer. Our published GPT-6 Astra evaluation scores 65.1 on the Coding Agent Index vs Codex’s 62.5, at 22% lower cost per solved task.
Free Options
Launch Team
Dropbox SignSimple, secure eSignatures for the way your team works.
Promoted
hey developers 👋
the problem
Coding agents write good code, but they’re still tied to your laptop. Close the lid and they die, your RAM caps how many you can run, and you still have to check every diff yourself.
the solution
Reason splits the agent in two. The brain (the harness) runs detached in the cloud and drives separate hands (a sandboxed machine). Sessions survive crashes, deploys and lost boxes, can be rewound, and hop between cloud Linux, a cloud Mac and your own Mac without losing state.
And it does more for less. On GPT-6 Astra, Reason scores 65.1 vs Codex’s 62.5 on the Coding Agent Index at 22% lower cost per solved task, and it’s 15–23% cheaper per attempt on every benchmark. On GPT-6.1 Sol, median cost per task is 34% below Codex. Every score links to a public Harbor job.
What’s inside
Built for the cache: a small set of tools and a short prompt, and each request only appends, so every turn is fully cache-eligible. That’s where the savings come from.
A computer per session: full desktop, live browser and terminals, with your saved logins. Take the wheel any time.
PRs with proof: walkthrough video and before/after screenshots go on every PR, then review, CI fixes and the merge queue.
Bring your own subs: plug in the coding plans you already pay for. When one hits its rate limit, Reason rolls over to the next.
Resume any agent: import a Claude Code, Codex, OpenCode or Devin session and keep going.
our ask
Try it at reasonmachines.com and tell us what broke: contact@reasonmachines.com
Hey Product Hunt! , I’m Adi, cofounder of Reason Machines.
I think a coding agent should finish useful work without needing an engineer to babysit it or burn through tokens.
A cost-efficient coding harness
We spend a lot of our time on the harness: the tools, context and execution loop around the model. We keep the toolset small, run tools in parallel and preserve the prompt cache.
In our published GPT-6 Astra evaluation, Reason achieved:
- 65.1 on the Coding Agent Index, versus Codex’s 62.5
- 22% lower cost per solved task
Same model, same tasks, same graders. [See the benchmarks and methodology](https://reasonmachines.com/blog/introducing-reason-agent-1-1).
In the cloud or on your computer
Give Reason a task and it writes code, runs tests and opens a pull request.
Use it from the Mac app, terminal, web or Slack, with your choice of models and connected subscriptions.
[Start free on your own computer](https://reasonmachines.com).
Try it on a task you’ve been putting off. Happy to answer questions about the product, harness or benchmarks.
Hey everyone , Lars here, one of the engineers building Reason 👋
Reason is an autonomous software engineer. Give it a task, and it gives you back a pull request you can review, with proof that the change works.
What a task looks like
1. Start from wherever you are: the Mac app, terminal, web, Slack, or your phone. Assign a GitHub issue or describe the work.
2. Reason gets its own computer for the session, with a terminal, live browser, and your repo. You can watch it work or take control at any time.
3. It writes the code, runs the tests, and uses the app to check the change the same way you would.
4. It opens a PR with a screen recording and before/after screenshots, then handles review comments, CI failures, and the merge queue.
Where it runs
Run it locally on your own machine for free, or in the cloud so it keeps going after you close your laptop. Use the models you like and the coding plans you already pay for. Access from all your devices, even when you running on your personal computer!
What we use it for every day
Fixing bugs with regression tests. Tracking down slow code. Building internal dashboards. Splitting big changes across subagents. If you’re in the middle of a Claude Code or Codex session, you can import it and let Reason finish the job.
Most of Reason is built by Reason, and that’s still how we decide what to fix next.
If you have a backlog ticket you keep putting off, give it to Reason and tell us how it went. I’ll be in the comments all day.
@larstalian For reference guys! This is Lars's Github in the last 2 weeks on Reason! https://github.com/asdf9384
The 22% lower cost sounds interesting . Does that hold up with larger codebases and longer tasks too?
@adrian_scott2 It holds pretty good! Reason handles a majority of its own codebase (2 million lines of code) across all products we have.
A mono repo with swift, typescript, rust and go well maintained.
The screen recording attached to a PR caught my eye. Seeing what the agent actually did in the app should help a reviewer understand the change before going through the diff.
@oleksii_sekundant Exactly, visual verification native to every PR where applicable.