specVersion: '0.1'
provider: tensor
name: Tensor — Rate Limits
description: |
  Tensor does not publish public rate-limit numbers. Each `x-tensor-api-key` is issued
  with a tier negotiated during Airtable onboarding, scaled to the partner's workload.
  Protocol-level concurrency is bounded by Solana mainnet itself: ~50,000 TPS theoretical,
  with per-account write-lock contention as the practical limit on hot collections.
limits:
- surface: REST
  scope: api-key
  rate:
    documented: false
    notes: "Negotiated per partner. Public reference: https://dev.tensor.trade/docs/authentication."
- surface: TX
  scope: api-key
  rate:
    documented: false
    notes: TX construction limits move with the REST tier; Solana RPC submission is downstream and not bounded by Tensor.
- surface: WebSocket
  scope: api-key
  rate:
    documented: false
    notes: Single connection per key recommended; multiplex subscriptions via subscribe/unsubscribe frames.
solana:
  consensus: Solana mainnet
  notes:
  - Blocks every ~400 ms; finality typically <2 s.
  - High-contention NFT mints may require priority fees; query GET /api/v1/priority-fees for the live recommendation.
backoff:
  strategy: exponential
  guidance:
  - Honor HTTP 429 with `Retry-After` headers where present.
  - Refresh `lastValidBlockHeight` from the TX response and resubmit before expiry.
sources:
- https://dev.tensor.trade/docs/authentication
- https://docs.tensor.trade/trade/api-and-sdk
updated: '2026-05-24'