apiCommonsRateLimits: "0.1" provider: name: Arcade.dev url: https://arcade.dev docs: https://docs.arcade.dev pricingUrl: https://arcade.dev/pricing apiBaseUrl: https://api.arcade.dev meta: created: '2026-05-22' modified: '2026-05-22' source: - https://arcade.dev/pricing - https://docs.arcade.dev/en/references/changelog note: > Arcade publishes plan-level quotas rather than per-second/per-minute request rates. The Engine's first-party MCP rate limits were "increased" in the 2026-03-27 changelog entry but specific request-rate numbers are not publicly documented. The plan-level quotas below are sourced from arcade.dev/pricing (verified 2026-05-22). quotas: - id: user-challenges description: Metered interactive consent events on hosted Arcade. plans: hobby: limit: 100 period: month overageRate: null currency: USD growth: limit: 600 period: month overageRate: 0.05 currency: USD enterprise: limit: custom period: custom - id: standard-tool-executions description: Standard tool execution count, all toolkits except those marked Pro. plans: hobby: limit: 1000 period: month overageRate: null currency: USD growth: limit: 2000 period: month overageRate: 0.01 currency: USD enterprise: limit: custom period: custom - id: pro-tool-executions description: Pro-tier tool execution count for higher-cost toolkits. plans: hobby: limit: 50 period: month overageRate: null currency: USD growth: limit: 100 period: month overageRate: 0.50 currency: USD enterprise: limit: custom period: custom - id: arcade-hosted-mcp-servers description: Concurrent Arcade-hosted MCP server instances kept running 24/7. plans: hobby: limit: 1 period: ongoing overageRate: null currency: USD growth: limit: null period: ongoing overageRate: 0.05 currency: USD unit: server-hour note: "Hosted MCP servers metered at $0.05 per server-hour beyond the included single server." enterprise: limit: custom period: custom requestRates: documented: false note: > Arcade does not currently publish per-endpoint requests-per-second limits on its public Engine API. The 2026-03-27 changelog records an unspecified increase to Arcade-managed first-party MCP rate limits. Self-hosted deployments are bounded by worker capacity rather than Engine-side throttling. headers: documented: false note: > The publicly published OpenAPI spec (https://api.arcade.dev/v1/swagger) does not declare standard `X-RateLimit-*` headers in its global responses block. retryPolicy: documented: partial note: > The Engine exposes retry-config endpoints under /v1/plugins; per-tool retry policies are configurable rather than a single fixed strategy.