Limits & Quotas

System limits, payload constraints, and concurrency quotas across Sayvy AI

Limits and Quotas

To ensure ultra-low latency (sub-500ms voice turn-around) and protect against infrastructure denial-of-service, Sayvy AI applies the following payload, concurrency, and operational limits.

---

Concurrency Quotas

Metric	Free Tier	Growth Tier	Enterprise Tier
Simultaneous Outbound Calls	2	25	Up to 1,000+
Simultaneous Inbound Trunks	2	50	Dedicated Trunks
Active Campaign Batches	1	10	Unlimited
Simultaneous WebRTC Rooms	5	100	Unlimited

Simultaneous call limits are evaluated across your entire workspace in real-time. If a dispatch exceeds your concurrency cap, the API returns a 429 concurrency_limit_reached error.

---

Payload Size Limits

Resource	Maximum Limit	Notes
JSON Request Body	10 MB	Applies to all REST endpoints.
Voice Agent System Prompt	32,000 characters	Maximum tokens depend on underlying LLM model.
Custom Function / Tool Declarations	64 tools per agent	Evaluated during conversation turn.
Audio Upload Size (Recordings / Samples)	50 MB	Supported formats: .wav, .mp3, .ogg, .flac.
Batch Campaign Contacts	10,000 per request	Use pagination or CSV upload for larger cohorts.

---

Operational Timeouts

Operation	Timeout	Action on Timeout
REST API Gateway	30 seconds	Returns 504 Gateway Timeout.
Voice Call Max Duration	2 hours	Automatically ends call session and triggers webhook.
Silence / Idle Timeout	120 seconds	Call terminated if recipient remains silent after 3 agent prompts.
Webhook Delivery Attempt	10 seconds	Retried 5 times with exponential backoff before being marked dead-letter.

---

Requesting Limit Increases

Enterprise customers can request increased concurrency quotas, custom TTS model instances, and private SIP trunking:

1. Visit the Billing & Plan tab in your Sayvy AI Console.
2. Click Request Quota Increase or reach out directly to your account manager at [e-mail].