Self-service capacity
For teams proving and operating a focused workload. Simple monthly execution capacity, no charge for protocol chatter.
HAPI MCP pricing
Keep your API estate AI-ready.
Do not create another software estate.
AI can generate MCP code. It does not remove the work of securing, deploying, testing, updating, and retiring every generated server. HAPI keeps your existing API contract at the center.
Talk about production capacityThe commercial model
Pay for capacity when you need it. License the API estate when it matters.
Enterprise API estate
For organizations standardizing how many production APIs stay available to AI. Capacity is contracted; API Units guide the conversation.
Production paths
Start without registration. Choose capacity when you are ready.
Not sure which production plan is right for you? We have you covered: start with the Pilot Accelerator, a focused 14-day proof before committing to a recurring plan.
Free
$0
no account required
36,000 MCP executions / month
20 per minute
Explore HAPI locally or in a low-volume environment.
- npm/Bun packages, native binaries, and Docker
- Community support
- A genuine production-shaped evaluation allowance
- On-prem or fully managed on the cloud
Starter
$99
per month
250,000 MCP executions / month
120 per minute
For a first production integration and small engineering teams.
- Commercial production capacity
- Selected production capabilities
- Standard support
- Fully managed on the cloud
- Production plugins (Open Telemetry, OAuth2, Webhooks, and more)
Pro
$399
per month
2 million MCP executions / month
600 per minute
For teams making API capabilities part of their product.
- Production capability set
- Priority support
- Capacity for sustained workloads
- Fully managed on the cloud
- Production plugins and Add-ons
Pilot Accelerator
one-time engagement
14-day proof
One workflow · one MCP Server
For teams that want evidence before choosing a production plan.
- Dev/test environment
- Basic governance
- 30 days of follow-up support
- A practical adoption recommendation
An execution is an MCP tools/call. Discovery and protocol operations are not billable capacity.
The ownership question
Generation is cheap. Ownership is not.
Open source and AI-generated code can reduce acquisition cost. They do not remove lifecycle ownership.
Build and own another MCP estate
- Implement and validate adapters
- Operate deployment, dependencies, and upgrades
- Keep API and MCP behavior synchronized
Keep APIs MCP-ready with HAPI
- Expose the API contract you already operate
- Apply a consistent MCP runtime model
- Scale the estate without pricing by tool count
Enterprise
Standardize AI access across your API estate.
Enterprise is not Pro with more executions. It covers estate licensing, contracted execution capacity, private deployment options, and the support needed to keep a growing API surface AI-ready.
API Unit: one production API/OpenAPI service, up to 50 operations. A 51–100 operation API uses two Units.
Estimate your API Units
Use production APIs only. This is an indicative estimate; an Enterprise agreement confirms the count from the actual API estate.
One API Unit covers up to 50 operations. For a varied estate, count each API from its own OpenAPI definition instead of relying on an average.
Zero-Shadow Pilot
Turn one proven workflow into an API-estate decision.
The Thin-Slice Pilot proves a focused workflow. For Enterprise teams, the Zero-Shadow Pilot carries that evidence into an API-estate adoption recommendation, including AI consumption, authorization, deployment, and the lifecycle difference between HAPI and shadow MCP code.
Production capacity conversation
Tell us what you are keeping AI-ready.
We use this to recommend the right capacity, Pilot Accelerator, or Enterprise approach.
Pricing details
Questions buyers ask
What is an MCP execution?
An MCP execution is a completed tools/call operation. Initialization, discovery, tool listing, health checks, prompts, and resource reads do not consume execution capacity.
What is an API Unit?
For Enterprise licensing, one API Unit covers one production API or OpenAPI service with up to roughly 50 operations. An API with 51–100 operations uses two Units. Count every production API separately, then add the Units together. This reflects the API estate kept AI-ready, not an arbitrary count of MCP tools.
Why not charge by MCP tool?
Tool design is an architecture decision. HAPI should help teams design useful AI capabilities, not reward them for combining or removing tools to lower an invoice.
What happens when capacity is reached?
Self-service plans are designed to prompt an upgrade before capacity becomes a problem. Enterprise agreements use contracted capacity, negotiated bursts, and agreed review terms rather than surprise usage bills.
Can I use HAPI without creating an account?
Yes. Free HAPI is available for experimentation and low-volume use through npm/Bun, native binaries, and Docker. A production conversation starts only when you want more capacity or commercial capabilities.