To ensure platform stability, fair resource allocation, and optimal performance, rate limits are enforced across all incoming requests. These limits apply universally whether you are interacting with our services via the standard API or through the Model Context Protocol (MCP).Rate limits are tracked and restricted on a per-user basis (x-api-key) and are categorized by the type of operation being performed (Read, Trading Write, or Staking Write). The system evaluates traffic across three distinct time windows to accommodate different usage patterns:Per-second (Burst): Designed to handle brief, sudden spikes in traffic without blocking requests.
Per-minute (Sustained): Governs steady, continuous traffic to prevent infrastructure strain.
Per-hour (AFC lever): Acts as a macro-level guardrail to limit total hourly resource consumption.
429 Too Many Requests HTTP status code until the rate limit resets.
Rate Limit Overview#
| Operation Category | Scope | Per-second (burst) | Per-minute (sustained) | Per-hour (AFC lever) |
|---|
| Read (prices, portfolio, history, asset browsing) | Per user | 20 req/s | 300 req/min | 3,000 req/hr |
| Trading Write (request quote, accept quote) | Per user | 5 req/s | 30 req/min | 100 req/hr |
| Staking Write (stake, unstake) | Per user | 5 req/s | 20 req/min | 60 req/hr |
Modified at 2026-07-01 08:13:34
