Pricing
Plan comparison
Compare plans side by side. Filter to the plans you're weighing, switch between monthly and annual billing, and see exactly what each tier includes.
Free
Test the full platform end-to-end with 30 included credits — no card required.
Choose FreeStarter
Hosted Stripe billing and prepaid credit packs when you need more headroom.
Choose StarterPro
Most popularPrivate registry, shared projects, and seat-based pricing for shipping real agents.
Choose ProMax
Higher limits, OAuth audit trails, and full audit history for scaled workspaces.
Choose MaxEnterprise
Dedicated infrastructure, self-hosted export, and governance for compliance and scale.
Talk to usEstimate your month
Workload → credits → spend
Credits meter the real cost of each action — build actions and served tool calls draw from the same pool. Drag to match a typical month and see which plan fits.
Credits bill the actual model cost of each action, so a quiet month costs less than a busy one — unlike flat per-call plans where you pay the same whether a call is a cheap read or an expensive generation. If your usage is steady and predictable, the table below shows the flat tier that lands cheapest.
| Capability | FreeFree | StarterStarter | ProPro | MaxMax | EnterpriseEnterprise |
|---|---|---|---|---|---|
| Usage limitsMonthly credits and resource ceilings per organization. | |||||
| Monthly included creditsRefreshes at the start of each calendar month (UTC). The Free tier receives a one-time welcome grant instead. | 30 welcome (one-time) | 150 | 500 | 2,000 | Custom |
| MCP servers | 2 | 10 | Unlimited | Unlimited | Unlimited |
| Skills | 3 | 15 | Unlimited | Unlimited | Unlimited |
| Active deployments | 1 | 5 | Unlimited | Unlimited | Unlimited |
| Requests per dayMCP requests served per caller, per deployed server, each day. Enforced at the edge; rate-limited rejections are never billed. | 100 | 1,000 | 10,000 | 50,000 | 250,000 |
| Burst ceilingRequests per minute per caller, on top of the daily allowance. Smooths spikes so one burst cannot spend a whole day of quota in seconds. | 30/min | 120/min | 600/min | 1,500/min | 3,000/min |
| Concurrent deploysHow many of your deploys build at once. Beyond this they queue in your lane rather than failing. | 1 | 1 | 2 | 3 | 4 |
| API keys | 3 | 10 | 25 | 100 | Unlimited |
| Registry & marketplaceWhere your servers and skills can be discovered and installed. | |||||
| Public registry browse & install | |||||
| Private organization registry | |||||
| Marketplace listingPublish to the public marketplace. Paid listings and creator payouts are coming soon. | |||||
| Server ingestionHow you bring API definitions and code into MCPCloud. | |||||
| OpenAPI / Swagger import | |||||
| GitHub repository import | |||||
| GraphQL schema import | |||||
| URL ingestion | |||||
| Code repository import | |||||
| DeploymentWhere your MCP servers can be deployed and exposed. | |||||
| Deploy queue priorityWhere your deploys sit when the platform is busy. Every deploy runs either way — priority changes when, not whether. | Standard | Standard | Priority lane | Front of line | Front of line |
| Cloudflare Workers deployment | |||||
| Environments per serverNamed deployment lanes beside production, each with its own hostname, upstream API and credentials — so a staging server runs alongside the live one. Credentials never cross between them. | 1 | 3 | 3 | Unlimited | |
| Premium subdomainDirect worker routing on a clean MCPCloud subdomain. | |||||
| Bring your own domainAttach your own domains and subdomains with automatic verification and managed TLS certs. | |||||
| Self-host exportDownload a runnable Docker bundle of any server and host it yourself. | |||||
| Code ModeThe model writes JavaScript against your tools and runs it in an isolated sandbox, so intermediate results never enter its context. | |||||
| Code Mode calls per executionHow many tool calls one sandboxed execution may make, within a 20-second execution window. The call ceiling bounds what a single runaway or injected execution can do; the window is what usually binds first, since each call is a round trip. Neither is a cost control — every call is metered the same as a direct one. | Up to 100, bounded by a 20s execution window | Up to 500, bounded by a 20s execution window | Up to 2,000, bounded by a 20s execution window | Up to 2,000, bounded by a 20s execution window | |
| SkillsSkill Studio authoring and runtime capabilities. | |||||
| Sandbox testing | |||||
| Skill composition (DAG) | |||||
| Auth & securityIdentity, access control, and network safeguards. | |||||
| OAuth provider | |||||
| Per-member OAuth connections | |||||
| Role-based permissions | |||||
| IP allowlisting | Planned | ||||
| SAML single sign-on | Planned | Planned | |||
| Collaboration & governanceShared workspaces and audit visibility. | |||||
| Shared organization projects | |||||
| Audit logs | Operations log | Org-level audit | Full audit history | Full audit history | |
| OAuth audit logs | |||||
| Billable membersPro and Max bill per seat; Free and Starter are single-user workspaces. | 1 | 1 | Per seat | Per seat | Custom |
| Billing & supportTop-ups, dedicated infrastructure, and support motion. | |||||
| Tool-call meteringSuccessful tool calls on deployed servers draw from the same credit pool as LLM usage. Failed, rate-limited, and protocol calls are never billed. | 0.15 credits / 1k calls | 0.15 credits / 1k calls | 0.15 credits / 1k calls | 0.15 credits / 1k calls | Committed rate |
| Credit pack purchases | |||||
| Usage overage capOpt-in headroom below a zero balance, billed back on the next invoice. | $20 | $100 | $500 | $2,000 | |
| Dedicated container runtimeRun a server on a dedicated container instance — full Node/Bun APIs, no isolate CPU caps. | Beta | Beta | |||
Marketplace creators keep 75% of revenue once paid listings launch. Talk to us about enterprise rollout, or email support.