Rate Limits & Quotas

What to expect per tier — predictable throughput for deterministic automation

⚡ AccelEx limits are based on completed business work units (invoices resolved, contracts processed, cases closed), not raw API calls. Each unit represents a deterministic, auditable outcome. Choose the tier that matches your operational volume.

📊 Per‑tier limits overview

Metric	Starter	Business	Enterprise	On‑Prem / Private
Work units / month	5,000	50,000	Custom (250k+)	Unlimited (licensed)
Work units / second (burst)	2	10	50+	Configurable
Concurrent pipeline executions	1	5	30	Based on infra
API requests per second (control plane)	5	20	100	Unlimited
Webhook delivery SLA	Best effort	99.9%	99.95%	99.99%
Audit log retention	30 days	90 days	1 year	Custom

⏱️ API rate limits (REST & GraphQL)

Headers you'll receive

X-RateLimit-Limit: 100
X-RateLimit-Remaining: 94
X-RateLimit-Reset: 1702944000
Retry-After: 12

Every response includes these headers. When limit exceeded, 429 Too Many Requests is returned with Retry-After (seconds).

Burst vs steady state

AccelEx uses token bucket algorithm. Short bursts allowed (e.g., 10 req/s for Business tier), but sustained throughput is enforced over 60‑second window. Webhook submissions and work unit polling share the quota.

🧠 Work unit quotas (core product)

What counts as a work unit? A completed, validated business outcome delivered from unstructured input. Examples:

✓ Invoice → resolved with approval signal
✓ Contract → fully parsed, risk‑scored, clause‑extracted
✓ Support case → summarized, categorized, routed
✓ Compliance record → structured and audited

Failed or low‑confidence workflows (manually routed) do not consume a unit until a final decision is produced. You only pay for resolved work.

📈 Rate limit per endpoint category

Endpoint group	Starter	Business	Enterprise
`POST /v1/workflows/start` (submit new unit)	2 req/s	10 req/s	50 req/s
`GET /v1/workflows/{id}/status`	5 req/s	20 req/s	100 req/s
`GET /v1/results/{id}` (retrieve outcome)	10 req/s	30 req/s	150 req/s
`POST /v1/webhooks/register`	1 req/s	3 req/s	10 req/s
`GET /v1/metrics` (dashboard APIs)	2 req/s	5 req/s	20 req/s

🔄 Quota management & burst handling

Exponential backoff (recommended)

async function callWithRetry(fn) {
  for (let i = 0; i < 5; i++) {
    try {
      return await fn();
    } catch (err) {
      if (err.status !== 429) throw err;
      const delay = Math.pow(2, i) * 1000;
      await new Promise(r => setTimeout(r, delay));
    }
  }
}

Quota increase requests

Business & Enterprise plans allow quota adjustments. Contact support@accelexpress.com or use Dashboard → "Request limit increase". For on‑prem deployments, limits are defined in the license contract with no hard caps.

✓ Enterprise includes priority rate limit lanes and dedicated throughput guarantees.

📌 Real‑world throughput examples

Starter tier: ~2 work units/second → ~172k units/day theoretical burst, but monthly cap 5k. Ideal for testing and low‑volume automation.
Business tier: 10 units/second sustained → up to 864k units/day, 50k/month quota (soft, can increase). Handles mid‑size invoice processing.
Enterprise: 50+ units/second, dynamic scaling. Used by Fortune 500 for real‑time contract ingestion and case resolution.

🚦Error handling & headers

429 response example

{
  "error": "rate_limit_exceeded",
  "message": "Work unit limit exceeded. Current: 5000/5000 (monthly). Resets at 2025-06-01T00:00:00Z",
  "retry_after": 86400,
  "limit_type": "monthly_quota"
}

Best practices

Implement idempotency keys to avoid duplicate submissions.
Monitor X-RateLimit-Remaining and X-RateLimit-Reset.
Use webhooks instead of polling for async results → reduces API calls.
Batch submissions when possible (bulk invoice endpoints have higher limits).

📋 FAQ — Rate limits & quotas

❓ What happens if I exceed my monthly work unit quota?
Additional submissions return 429 until the next billing cycle or quota top‑up. You can purchase overage packs or upgrade tier via Dashboard.

❓ Do failed workflows count against my limit?
No. Only completed, validated business outcomes count. If a workflow returns confidence_score below threshold or fails validation, no unit is deducted.

❓ Can I pool quotas across multiple pipelines?
Yes, quota is shared across all pipelines (finance, legal, support) within a tenant. Enterprise supports dedicated pools per pipeline.

❓ Are webhook deliveries rate‑limited?
Outbound webhooks from AccelEx to your endpoint respect your server's responsiveness. If your endpoint slows down, we throttle delivery automatically (no penalty to your quota).

❓ How to monitor real‑time usage?
Dashboard provides live graphs. Also GET /v1/usage/current returns remaining quota and reset timestamps.

⚙️ Need higher burst or custom SLA? Contact our team for dedicated rate limit tiers or on‑premise deployment with no artificial constraints.

Compare plans → Request custom limits