Rate Limits & Quotas

Understand API usage constraints, tier-specific allocations, throttle behavior, and how to monitor or request increased limits for your NexusAI integration.

Default Rate Limits

Rate limits are applied per API key and are calculated using a sliding window algorithm. Limits vary based on your subscription tier.

Limit Type	Starter	Professional	Enterprise
Requests / Minute	60	600	Unlimited*
Requests / Day	10,000	500,000	Custom
Concurrent Connections	5	25	100+
Max Payload Size	4 MB	16 MB	100 MB
Custom Model Slots	1	5	Unlimited

Note: Enterprise limits are tailored to your workload. Contact your solutions engineer to define custom thresholds. Burst allowances of +20% are permitted during peak windows. Burst allowances of +20% are permitted during peak windows.

Rate Limit Headers

Every API response includes headers to help you track consumption and prevent throttling.

                    X-RateLimit-Limit: 600
                    X-RateLimit-Remaining: 487
                    X-RateLimit-Reset: 1718923456 (Unix timestamp)
                    Retry-After: 45 (Seconds, only on 429)
                

Monitor Remaining in your client logic. If it drops below 10%, scale back or queue requests.

Throttling & Retry Behavior

When you exceed your limit, NexusAI returns a 429 Too Many Requests status. The response body includes a structured error:

                    {
                      "error": {
                        "code": "RATE_LIMIT_EXCEEDED",
                        "message": "You have exceeded the allowed request rate.",
                        "retry_after": 30
                      }
                    }
                

Recommended Strategy

Implement exponential backoff with jitter: delay = min(2^attempt * 1000 + random(0, 1000), 30000)
Never retry immediately on 429. Respect the Retry-After header or retry_after field.
Use connection pooling and batch endpoints where available to reduce overhead.

Warning: Aggressive retrying may result in temporary IP flagging. Consistent 429s can trigger automated cooldowns. Design idempotent requests to safely retry without side effects. Consistent 429s can trigger automated cooldowns. Design idempotent requests to safely retry without side effects.

Monitoring Usage

Usage API

Fetch real-time and historical consumption data programmatically:

                    GET /v1/usage

                    Query params: start, end, resolution (hour|day|month)

                    Returns: JSON object with total_requests, quota_used, quota_limit, breakdown

Dashboard & Alerts

Log into the NexusAI Console to view graphical usage trends, set threshold alerts (50%, 80%, 95%), and receive notifications via email, Slack, or webhook.

Requesting Quota Increases

Need higher limits for a campaign, model training job, or seasonal spike?

Navigate to Settings > API Keys > Quota Requests in the console.
Specify the endpoint, desired limit, duration, and business justification.
Enterprise plans can request permanent increases; others receive temporary boosts (max 30 days).

Approvals typically take 1–2 business hours. Temporary limits automatically revert after the specified window.

Frequently Asked Questions

What happens if I exceed my daily quota?

Requests will be rejected with a 429 until the quota resets at 00:00 UTC. No overage charges are applied.

Are limits shared across team members?

Yes. All API keys under a single organization share the same pool. Use key-based tracking to monitor individual service consumption.

Can I upgrade limits instantly?

Upgrading your plan applies new limits immediately. Temporary boosts require manual approval but are typically granted within the hour.

How are streaming endpoints counted?

Streaming calls count as a single request initiation. Each chunk is not counted against the rate limit, but total duration affects concurrent connection limits.

← Authentication | Error Handling →