// PRICING
Deploy instantly.
Scale infinitely.
From proof of concept to enterprise. Including a free option.
No credit card. No hidden fees. No per-seat pricing.
Starting from
/month
billed annually — $288
Sell on the Marketplace. No plan needed.
List your MCP server and earn — without a subscription.
Included in all plans
// FREE
Proof of concept. See it work before you commit.
3 MCP Servers
1 Connection Token
10 MCP Requests / mo
Included when you sign up.
// DEVELOPER
Individuals & small projects exploring AI integrations
Unlimited Servers
Unlimited Deploys
Up to 3 Connection Tokens
10,000 MCP Requests / mo
Billed annually at $288
// TEAM
Growing teams with production systems
Unlimited Servers
Unlimited Deploys
Up to 10 Connection Tokens
100,000 MCP Requests / mo
Billed annually at $852
// BUSINESS
Teams shipping AI to production with governance
Unlimited Servers
Unlimited Deploys
Up to 25 Connection Tokens
500,000 MCP Requests / mo
Billed annually at $1,908
// BILLING
Honest billing. No surprises.
MCP runs on persistent SSE infrastructure. That's why compute has a real cost — and why we show you exactly what you pay.
MCP Request
One request from an AI client to your MCP server. One tool call, one prompt resolution, one action.
Cursor calling your server to list tasks = 1 MCP Request
Connection Token
A persistent authenticated channel between one AI client and your MCP server. One token = one active connection.
Claude Desktop + your server = 1 Connection Token
Soft Limit
You set the pace. We never cut you off.
Never interrupted. Never cut off.
$5 per 10,000 additional MCP Requests
SSE keeps a persistent connection open between your AI client and your server. That's what makes it fast — and what makes compute cost real.
No per-seat pricing. No plan needed to sell. Ever.
Frequently Asked Questions
Everything you need to know about Vinkius pricing.
How are MCP Requests counted?
Requests are pooled at the account level. One request is counted each time an AI tool queries your API through Vinkius. All your Connection Tokens share the same pool. For example, the Team plan includes 100,000 requests shared across all tokens.
What happens if we exceed the monthly limit?
Nothing breaks. Your AI tools keep working without interruption. We use a soft limit — if you go over, the extra usage is billed at $5 per 10,000 additional requests. We never cut your connection.
What is a Connection Token?
A Connection Token is a secure key that connects a specific AI client (Cursor, Claude, VS Code, etc.) to your MCP server. Your real API credentials stay encrypted — you only share scoped tokens. Each plan has a token limit per server: Developer gets 3, Team gets 10, Business gets 25. If someone leaves or a device is compromised, revoke their token — it's disconnected globally in under 40ms.
What does the AI cost optimizer (FinOps) do?
APIs often return far more data than AI actually needs. The basic plan trims oversized responses to keep things working. The FinOps optimizer (Team plan and above) goes further — it intelligently reduces data to only what's relevant, strips empty fields, and limits arrays to your specifications. Result: faster responses, lower AI costs, better accuracy.
How does privacy protection (DLP) work?
AI needs access to real data, but sensitive information must stay protected. Vinkius automatically detects and hides sensitive data (credit cards, social security numbers, emails) before it reaches the AI. This happens entirely in memory — nothing is saved to disk. Your data stays compliant by default.
What is Read-Only Safe Mode?
AI can sometimes take unexpected actions. Safe Mode prevents this by hiding all write operations (create, update, delete) from the AI. The AI can only read data — it physically cannot see or execute any action that modifies your systems.
How does the instant kill-switch work?
When someone leaves your team or a device is compromised, changing an API key isn't enough — active connections stay open. With Vinkius, click 'Revoke' and the connection is force-closed globally in under 40 milliseconds. Every other connection stays untouched.
How does SSO and access management work?
In DevelopmentSSO and advanced access management are currently in active development. It will integrate with your corporate identity provider (Okta, Entra ID, Google Workspace). Your existing access rules remain the single source of truth — Vinkius respects your organization's permissions.
What is your uptime guarantee?
We maintain 99.9% uptime across all plans. Our infrastructure runs on globally distributed servers with automatic failover. If we ever fall below our SLA, affected customers receive service credits.
What is the refund policy?
We do not offer refunds. You can cancel anytime. We recommend starting with the free tier (1 token, 10 requests) to test the platform before subscribing.
Do I need a plan to sell on the Marketplace?
No. Listing and selling on the Vinkius Marketplace is free — forever. No subscription required. A compute fee is deducted from your revenue, never charged separately.
Is the Free tier really free forever?
Yes. $0 per month, no credit card, no time limit. It includes 10 MCP Requests per month — enough to verify the product works for your use case. Not designed for production workloads.
What happens when I hit my request limit?
Nothing breaks. Your AI tools keep working without interruption. We use a soft limit — if you go over, the extra usage is billed at $5 per 10,000 additional requests. We never cut your connection.
Can I downgrade or cancel?
Yes, at any time. No lock-in, no penalties. A downgrade takes effect at the end of your current billing cycle. You keep full access until then.