Theme
Palette
Your own colour
Saved for this browser, on every page.
Pricing
Start free. Move up when your runs or your team outgrow the plan. Calls on your own provider key are billed by that provider, not by us.
Plans
Every plan includes agents, workflows, tools, traces and run history. What changes is how much you may run and how many people share it.
USD 0
Build and test a small AI product.
USD 19
Deploy AI workflows for one product.
USD 49
Operate production workflows across environments.
USD 149
Run governed AI workflows with a larger team.
Custom
Use custom limits, security, and support terms.
Managed usage
Top up managed model usage for a busy month without changing plan. Credits apply to calls that run through Vendor.sh, never to calls on your own provider key.
Starter credits
Add USD 10 of managed AI usage value.
USD 10
Growth credits
Add USD 50 of managed AI usage value.
USD 50
Scale credits
Add USD 100 of managed AI usage value.
USD 100
Compare
| Every limit the runtime checks. | Free | Builder | Pro Recommended | Team | Enterprise |
|---|---|---|---|---|---|
| Usage | |||||
| Monthly runs | 200 | 5,000 | 25,000 | 100,000 | Unlimited |
| Concurrent runs | 1 | 3 | 10 | 25 | Unlimited |
| Run retention | 7 days | 30 days | 90 days | 180 days | Unlimited |
| Workspace | |||||
| Seats | 1 | 1 | 5 | 15 | Unlimited |
| Projects | 1 | 5 | Unlimited | Unlimited | Unlimited |
| Active workflows | 2 | 10 | 50 | Unlimited | Unlimited |
| Environments | 1 | 1 | 3 | 3 | Unlimited |
| Models | |||||
| Bring your own provider key | Included | Included | Included | Included | Included |
| Vendor.sh managed credits | Included | Included | Included | Included | Included |
| Build and release | |||||
| Typed workflow API | Included | Included | Included | Included | Included |
| Signed webhooks | Included | Included | Included | Included | Included |
| Schedules | Not included | Included | Included | Included | Included |
| MCP client | Not included | Included | Included | Included | Included |
| MCP server | Not included | Not included | Included | Included | Included |
| Workflow evaluations | Not included | Not included | Included | Included | Included |
| Approval policies | Not included | Not included | Not included | Included | Included |
| Shared connections | Not included | Not included | Not included | Included | Included |
| Organisation | |||||
| Advanced workspace roles | Not included | Not included | Not included | Included | Included |
| Audit logs | Not included | Not included | Not included | Included | Included |
| Advanced run logs | Not included | Not included | Included | Included | Included |
Billing questions
One execution of a deployed workflow or agent, from the request that starts it to its final status. The steps inside it do not count separately.
A run is refused rather than queued indefinitely, and the refusal says which limit was reached. Raise the plan, or add a credit pack for managed usage.
The platform execution counts against your plan. The model usage does not: a provider call served by your key creates no managed provider debit here, and spends no credits.
Managed model or paid tool usage, plus the disclosed service amount. They are not spent on a call served by your own key.
No. A request rejected before scheduling does not consume a platform execution.
Yes, at any time. The workspace keeps its agents, workflows, runs and records; what stops is the paid allowance.
Start on Free, connect your own key, and move up when your runs outgrow it.