Claude Code and Codex, running in our cloud. Mention @Computer in Slack, or call them from your code with the API.
Bring your own subscription:Claude Pro or MaxChatGPT Plus or Proor an API keyNo subscription? Use ours at cost.
Benchmark
As fast as Claude Code on your laptop, and twice as fast as other cloud coding agents on the same task.
Read the benchmark →It’s fast. Really, really fast.
It answers while the question is still on your mind.
Dana4:12 PM
ComputerAPP4:12 PM
✓ Read PricingCard.tsx · Edit +6 −3 · Screenshot
Dana4:12 PM
ComputerAPP4:13 PM
✓ Edit +1 −1 · Screenshot
Dana4:13 PM
ComputerAPP4:13 PM
● Checks running
For engineers
Small changes that follow the patterns already in your repo. You review and merge.
Utpal10:02 AM
ComputerAPP10:02 AM
Splitting this three ways:
ComputerAPP10:04 AM
✓ Review: commas in customer names broke columns. Quoted, test added.
Add CSV export to invoices#248
What changed, why, and how it was checked. Links back to this thread.
✓ lint✓ typecheck✓ tests✓ preview+184 −9
For the team
Everyone in the thread sees the work and can reply to it.
Mira2:31 PM
ComputerAPP2:31 PM
✓ Edit src/onboarding/steps.ts
Moved it to the end. Preview is up on the branch.
Igor2:40 PM
Utpal9:05 AM
ComputerAPP9:06 AM
✓ Edit src/emails/inviteReminder.tsx
Added on the same branch, so it only goes to teams that used Igor's skip link.
Pricing
Try it free for a week. Then pick a plan for your whole Slack workspace: you pay for work minutes only. Bring your own model, or use ours at cost. We're fast, so your minutes go further.
Your model
If your plan hits its limit
@Computer keeps working on our tokens at API prices, no markup, and switches back to your subscription when its limit resets. We tell you in the thread when it switches.
Your team works
2,000 work minutes a month
You pay
$20 / month
$0 for the model, on your Claude or ChatGPT plan
Pro, 2,000 minutes included
Plans cover the whole workspace. Past your plan's minutes, each extra minute is $0.01, paid from prepaid credits: packs of $10, $50, $200, with optional auto top-up. Nothing is billed after the fact. Models are yours at no charge from us, or ours at API prices with no markup, also from credits.
The clock runs while it works
From the moment @Computer picks up your message to its reply: reading code, running commands, builds and tests. Metered by the second.
Waiting is free
Time between messages, idle threads and time spent waiting on you to review costs nothing. Two threads working at once count as two. Your serverless agents draw on the same minutes.
Why minutes go further
Workspaces resume warm. Your repo, dependencies and build caches are already there, so a minute goes on your work, not on cloning and reinstalling.
Serverless agents
Write an agent as a TypeScript function and deploy it with one command. Run it on a schedule, or call it from your product over HTTP.
01 · Deploy it
import { useModel, useTool } from "@opencomputer/agent";
import { openCleanupPullRequest } from "./tools/github.js";
export default function Agent() {
useModel("anthropic/claude-sonnet-4.6");
useTool(openCleanupPullRequest);
return "Find stale feature flags, open a PR to remove each.";
}02 · Call it as an API
API=https://app.opencomputer.dev/api/managed-agents
# start a session with your deployed agent
curl -X POST $API/sessions -H "x-api-key: $KEY" \
-H "content-type: application/json" \
-d '{"agentId": "flag-cleaner@production"}'
# send it work
curl -X POST $API/sessions/$ID/turns -H "x-api-key: $KEY" \
-H "content-type: application/json" \
-d '{"input": "Clean up stale flags in billing."}'Any model
Pick the model per agent, or bring your own key.
Keys stay outside
Your API keys and secrets never enter the runtime.
Built for long work
Durable sessions, schedules and webhooks are included.
Serverless agents share your plan's work minutes with @Computer. See pricing
Add it to Slack, then mention @Computer in any channel.