Pricing

Bring your own key. Pay per agent. Cap the spend.

No markup on inference and no per-seat tax. Connect your own model key, pay only for the agents that actually run, and set a hard monthly budget cap that pauses new agents at the ceiling. There's no payment integration yet — join the waitlist for early access.

Solo

For a founder or a single builder steering a couple of agents.

Free core

self-host the open-core, bring your own key

  • Bring your own model key (DeepSeek, Claude, Kimi, OpenAI)
  • Up to 2 concurrent agents
  • Hard monthly budget cap — spawns pause at the ceiling
  • Full transcript, idea intake, branch-per-agent builds
  • Community support
Get early access

Team

Most popular

For a product engineering team watching and steering a fleet together.

Per agent

watchers are free — you pay for running agents, not seats

  • Everything in Solo
  • Priced per running agent — invite the whole team to watch for free
  • Higher concurrency caps, per-workspace budget caps
  • Human-approved merges with shared review
  • Managed hosting option + email support
Get early access

Enterprise

For orgs that need SSO, self-hosting, and per-team budget governance.

Let's talk

volume agent pricing, annual, and a commercial EE license

  • Everything in Team
  • Self-hosted or private deployment (commercial EE)
  • SSO / SAML and audit logging
  • Org-wide and per-team budget caps
  • Priority support and onboarding
Contact sales
How pricing works. BYOK means you pay your model provider directly at cost — we never mark up inference. On top of that, OurAI meters the agents that run, so watchers are always free. A budget guard tracks spend across running agents and pauses new spawns the moment you hit your monthly cap.

Frequently asked

What does bring-your-own-key (BYOK) mean?

You connect your own model provider key — DeepSeek, Claude, Kimi, or OpenAI — through the pluggable model gateway. Model tokens are billed to you by that provider at cost. OurAI never marks up inference; we meter the agents that run on top.

Why price per agent instead of per seat?

Cost should scale with the compute doing the work, not your headcount. Anyone on the team can watch and steer a live session for free; you only pay for the agents actually running. Add reviewers without adding to the bill.

How do the monthly budget caps work?

You set a hard ceiling in USD. A budget guard tracks spend across running agents and pauses new agent spawns once you reach the cap — no surprise overages. Raise or lower the cap whenever you need to.

Is there a free tier?

The open-core is free to self-host, and Solo runs on your own key. Join the early-access waitlist to get in — no credit card required, because there is no payment integration yet.

Can I self-host?

Yes. The core is open (AGPL-3.0) and the SDK is Apache-2.0. Enterprise adds a commercial EE license for private and self-hosted deployments with SSO and governance.

No card, no per-seat tax. Bring your team and your key.

Get early access