Anthropic AI Gateway Benchmark
Request Latency
Cold E2E
738ms
Median
Warm TTFT
595ms
Median
Tokens/sec
50
Median
DNS
13.0ms
Median
TCP
1.6ms
Median
TLS
3.5ms
Median
Cold Connection Breakdown
Where a fresh request's time goes: DNS lookup, TCP connect, TLS handshake, then time to first response byte, then first byte to first streamed token.
DNS13.0ms
TCP1.6ms
TLS3.5ms
To First Byte690.4ms
First Byte → Token0.2ms
Performance Over Time
Iteration Distribution
Cold E2E
Warm TTFT
Tokens/sec
DNS
TCP
TLS
Become a partner
Want your logo featured across our benchmarks? Become a ComputeSDK benchmark partner.