Charts · Gateways · On the Bench · AI tool benchmarks

Every chart here is drawn from the same numbers the tables rank, on the cell and tail-latency bound last chosen on the table tabs; the line under each chart names them. Bars are coloured by implementation language (key above the chart), a neutral property, never by rank or brand.

  • Rust
  • Other
  • Go
  • Node
  • Python
CPU per request by gateway
RankGatewayCPU per request
1Busbar69 µs
2AISIX (api7)153 µs
3APISIX186 µs
4agentgateway189 µs
5Kong221 µs
6Helicone259 µs
7LiteLLM · Rust273 µs
8GoModel439 µs
9TensorZero459 µs
10Higress709 µs
11Bifrost865 µs
12Portkey1.5 ms
13LiteLLM · Python8.6 ms
14Plano182.2 ms

Microseconds of gateway CPU per completed request, measured at one concurrency held identical for every gateway. Lower is better. Read on each gateway's own representative same-dialect cell, as last chosen on the table tabs. Drawn on a log scale: this metric spans orders of magnitude, and a linear axis would collapse most of the field into the width of a line.
Not shown (no value on the chosen cell): One-API.