Benchmarks
Real, reproducible performance data for Telnyx AI and communications infrastructure. Every result links to its methodology so you can verify it yourself.
6/6
Passing
| Benchmark | Result | Status | Direction |
|---|---|---|---|
p99 API Latency 99th percentile round-trip time for API requests from US-East | 47ms | ✓ Pass | ↓ Lower is better |
Vendor Count for Voice AI Agent Distinct vendors required for a production voice AI agent | 1 vendor | ✓ Pass | ↓ Lower is better |
Carrier Network Ownership Whether the provider owns its carrier network infrastructure | Carrier-owned | ✓ Pass | ↑ Higher is better |
STIR/SHAKEN Attestation Level Call authentication attestation level (A=highest, C=lowest) | A-level | ✓ Pass | ↑ Higher is better |
SIP Outbound Call Setup Latency Time from SIP INVITE to 200 OK for outbound calls | 85ms | ✓ Pass | ↓ Lower is better |
Global Datacenter Count Number of owned datacenter locations | 4 (2 US, 1 EU, 1 APAC) | ✓ Pass | ↑ Higher is better |
Reproducible
Every benchmark includes a CLI command you can run to independently verify the results.
npx @telnyx/benchmarkHow each benchmark is measured, what counts as passing, and how to reproduce results.
Machine-readable benchmark data for automated evaluation and agent-driven workflows.
