Skip to main content
Metabind bills on four simple axes — developer seats, tool invocations, bandwidth, and asset storage. LLM tokens run on your own model key, billed by your model provider with no Metabind markup, on every tier. The Assistant SDK is included everywhere.

Plans

Free

$0/monthShip your first MCP App.
  • 1 developer seat
  • 3 MCP Apps
  • 10K tool invocations
  • 10 GB bandwidth
  • 5 GB storage
  • Your own model key for LLM tokens

Developer

$49/monthor $529/year — save 10%For developers shipping MCP Apps.
  • 1 developer seat (+$99 each)
  • 5 MCP Apps
  • 100K tool invocations
  • 100 GB bandwidth
  • 50 GB storage
  • Tool-call analytics + email support

Team

$499/monthor $5,389/year — save 10%For teams scaling across projects.
  • 5 developer seats (+$99 each)
  • 25 MCP Apps
  • 1M tool invocations
  • 1 TB bandwidth
  • 250 GB storage
  • RBAC, audit logs, priority support

Enterprise

Custom pricingFor organizations with custom requirements.
  • Custom seats & usage
  • Unlimited MCP Apps
  • SSO/SAML
  • Audit logs (1-year retention)
  • Custom SLA
  • Dedicated success manager & priority support

What counts as a seat?

Seats are authoring access to Metabind Studio — the people who build components, define tools, and publish. Ship to unlimited end-users with no per-user fees. You pay only for authoring seats and usage.

What’s included in every plan

  • Metabind Studio — visual component builder, code editor, and live device previews
  • BindJS — declarative authoring layer, Apache 2.0 (60+ components, 40+ modifiers)
  • Schema-validated Interactive Tools — governed UI, validated on every render
  • Data Tools in V8 isolates — wrap any REST or GraphQL API; secrets injected at runtime
  • One-button publish — live across Claude, ChatGPT, VS Code, and every MCP host
  • Assistant SDK — run your agent inside your own iOS, Android, or web app
  • Native rendering in your own apps — SwiftUI, Jetpack Compose, and React from a single definition
  • Package versioning + instant rollback
  • Your own model key — Anthropic, OpenAI, or Gemini, billed by your provider with no markup

Usage & overages

Both tool types — Interactive Tools and Data Tools — count as tool invocations through the same metered compute path. Data Tool executions that run longer than 30 seconds count as 5 invocations. If you exceed your plan limits, usage is billed at these rates: Save 10% on any paid plan with annual billing.

Feature comparison

Platform

Custom domains

Governance & security

Analytics & support

Get started

Create Free Account

Start building for free

Contact Sales

Discuss enterprise requirements