Bifrost benchmark measurements
Operated by the Busbar team; Busbar is also an entrant. Gateway overhead against a local mock is not end-to-end model latency. Methodology and limitations.
Go · recorded build: maximhq/bifrost:v2.5.0. Measured: .
JSON · Markdown · CSV · Project source · Interactive dashboard
Each row below names its protocol pair. A representative dashboard cell may differ between gateways; compare an identical pair before interpreting a ranking. Unavailable metrics retain their reason; below-resolution readings are not exact zeroes.
Published measurements
| Ingress → upstream | Metric | Value | State / reason | p99 bound (ms) |
|---|---|---|---|---|
| OpenAI → OpenAI | Added latency p50 (µs) | 658 | Valid · n=8,543 gateway / 269,310 direct | — |
| OpenAI → OpenAI | Added latency p99 (µs) | 1,057 | Valid · n=8,543 gateway / 269,310 direct | — |
| OpenAI → OpenAI | Gateway p99 at concurrency 1 (µs) | 1,084 | Valid · n=8,543 | — |
| OpenAI → OpenAI | Direct-to-mock p99 at concurrency 1 (µs) | 27 | Valid · n=269,310 | — |
| OpenAI → OpenAI | CPU per request (µs) | 865.1 | Valid | — |
| OpenAI → OpenAI | Requests per CPU-second | 1,155.9 | Valid | — |
| OpenAI → OpenAI | Cost window concurrency | 8 | Valid | — |
| OpenAI → OpenAI | Cost window core utilisation | 0.991 | Valid | — |
| OpenAI → OpenAI | Cost window successful requests | 27,580 | Valid | — |
| OpenAI → OpenAI | Cost window throughput (req/s) | 4,595 | Valid | — |
| OpenAI → OpenAI | Cost window threads | 13 | Valid | — |
| OpenAI → OpenAI | Non-voluntary context switches per request | 0.027 | Valid | — |
| OpenAI → OpenAI | Cost window major page faults | 0 | Valid | — |
| OpenAI → OpenAI | Added time to first event p50 (µs) | 932 | Valid · n=100 gateway / 100 direct | — |
| OpenAI → OpenAI | Added time to first event p99 (µs) | 1,417 | Valid · n=100 gateway / 100 direct | — |
| OpenAI → OpenAI | Added inter-frame gap p50 (µs) | 3 | Valid · n=63 gateway / 63 direct | — |
| OpenAI → OpenAI | Added inter-frame gap p99 (µs) | — | Below resolution · n=63 gateway / 63 direct | — |
| OpenAI → OpenAI | Streams sustained (frames/s) | 25,840.9 | Valid | — |
| OpenAI → OpenAI | Streams sustained | 580 | Valid | — |
| OpenAI → OpenAI | Idle RSS (MiB) | 160.3 | Valid | — |
| OpenAI → OpenAI | Steady-state RSS (MiB) | 1,049.5 | Valid | — |
| OpenAI → OpenAI | Recovered RSS (MiB) | 1,256.6 | Valid | — |
| OpenAI → OpenAI | Peak RSS (MiB) | 1,257 | Valid | — |
| OpenAI → OpenAI | Peak RSS high-water mark (MiB) | 1,137.4 | Valid | — |
| OpenAI → OpenAI | Time to plateau (s) | — | Not measured | — |
| OpenAI → OpenAI | Memory growth (MiB/min) | 56.82 | Valid | — |
| OpenAI → OpenAI | Memory load duration (s) | 300 | Valid | — |
| OpenAI → OpenAI | Memory shape under load | 1 | Valid | — |
| OpenAI → OpenAI | Idle memory shape | — | Not measured | — |
| OpenAI → OpenAI | Idle memory static | 1 | Valid | — |
| OpenAI → OpenAI | Idle memory growth (MiB/min) | 0.48 | Valid | — |
| OpenAI → OpenAI | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| OpenAI → OpenAI | Throughput (req/s) | 4,346 | concurrency 4; n=26,081 | 5 ms |
| OpenAI → OpenAI | Throughput (req/s) | 4,555 | concurrency 8; n=27,340 | 10 ms |
| OpenAI → OpenAI | Throughput (req/s) | 4,624 | concurrency 16; n=27,748 | 50 ms |
| OpenAI → OpenAI | Throughput (req/s) | 4,624 | concurrency 16; n=27,748 | 100 ms |
| OpenAI → OpenAI | Throughput (req/s) | 4,624 | concurrency 16; n=27,748 | Unbounded |
| OpenAI Responses → OpenAI | Added latency p50 (µs) | 586 | Valid · n=9,572 gateway / 263,867 direct | — |
| OpenAI Responses → OpenAI | Added latency p99 (µs) | 944 | Valid · n=9,572 gateway / 263,867 direct | — |
| OpenAI Responses → OpenAI | Gateway p99 at concurrency 1 (µs) | 971 | Valid · n=9,572 | — |
| OpenAI Responses → OpenAI | Direct-to-mock p99 at concurrency 1 (µs) | 27 | Valid · n=263,867 | — |
| OpenAI Responses → OpenAI | CPU per request (µs) | 842.7 | Valid | — |
| OpenAI Responses → OpenAI | Requests per CPU-second | 1,186.7 | Valid | — |
| OpenAI Responses → OpenAI | Cost window concurrency | 8 | Valid | — |
| OpenAI Responses → OpenAI | Cost window core utilisation | 0.993 | Valid | — |
| OpenAI Responses → OpenAI | Cost window successful requests | 28,386 | Valid | — |
| OpenAI Responses → OpenAI | Cost window throughput (req/s) | 4,730 | Valid | — |
| OpenAI Responses → OpenAI | Cost window threads | 13 | Valid | — |
| OpenAI Responses → OpenAI | Non-voluntary context switches per request | 0.005 | Valid | — |
| OpenAI Responses → OpenAI | Cost window major page faults | 0 | Valid | — |
| OpenAI Responses → OpenAI | Added time to first event p50 (µs) | — | Untestable | — |
| OpenAI Responses → OpenAI | Added time to first event p99 (µs) | — | Untestable | — |
| OpenAI Responses → OpenAI | Added inter-frame gap p50 (µs) | — | Untestable | — |
| OpenAI Responses → OpenAI | Added inter-frame gap p99 (µs) | — | Untestable | — |
| OpenAI Responses → OpenAI | Streams sustained (frames/s) | — | Untestable | — |
| OpenAI Responses → OpenAI | Streams sustained | — | Untestable | — |
| OpenAI Responses → OpenAI | Idle RSS (MiB) | 161.2 | Valid | — |
| OpenAI Responses → OpenAI | Steady-state RSS (MiB) | 1,095.2 | Valid | — |
| OpenAI Responses → OpenAI | Recovered RSS (MiB) | 1,073.7 | Valid | — |
| OpenAI Responses → OpenAI | Peak RSS (MiB) | 1,252.7 | Valid | — |
| OpenAI Responses → OpenAI | Peak RSS high-water mark (MiB) | 1,167.7 | Valid | — |
| OpenAI Responses → OpenAI | Time to plateau (s) | — | Not measured | — |
| OpenAI Responses → OpenAI | Memory growth (MiB/min) | 73.11 | Valid | — |
| OpenAI Responses → OpenAI | Memory load duration (s) | 300 | Valid | — |
| OpenAI Responses → OpenAI | Memory shape under load | 1 | Valid | — |
| OpenAI Responses → OpenAI | Idle memory shape | — | Not measured | — |
| OpenAI Responses → OpenAI | Idle memory static | 1 | Valid | — |
| OpenAI Responses → OpenAI | Idle memory growth (MiB/min) | 0.355 | Valid | — |
| OpenAI Responses → OpenAI | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| OpenAI Responses → OpenAI | Throughput (req/s) | 4,373 | concurrency 4; n=26,247 | 5 ms |
| OpenAI Responses → OpenAI | Throughput (req/s) | 4,576 | concurrency 8; n=27,465 | 10 ms |
| OpenAI Responses → OpenAI | Throughput (req/s) | 4,629 | concurrency 16; n=27,782 | 50 ms |
| OpenAI Responses → OpenAI | Throughput (req/s) | 4,629 | concurrency 16; n=27,782 | 100 ms |
| OpenAI Responses → OpenAI | Throughput (req/s) | 4,629 | concurrency 16; n=27,782 | Unbounded |
| Anthropic → OpenAI | Added latency p50 (µs) | 597 | Valid · n=9,413 gateway / 262,256 direct | — |
| Anthropic → OpenAI | Added latency p99 (µs) | 968 | Valid · n=9,413 gateway / 262,256 direct | — |
| Anthropic → OpenAI | Gateway p99 at concurrency 1 (µs) | 996 | Valid · n=9,413 | — |
| Anthropic → OpenAI | Direct-to-mock p99 at concurrency 1 (µs) | 28 | Valid · n=262,256 | — |
| Anthropic → OpenAI | CPU per request (µs) | 863.6 | Valid | — |
| Anthropic → OpenAI | Requests per CPU-second | 1,157.9 | Valid | — |
| Anthropic → OpenAI | Cost window concurrency | 8 | Valid | — |
| Anthropic → OpenAI | Cost window core utilisation | 0.991 | Valid | — |
| Anthropic → OpenAI | Cost window successful requests | 27,627 | Valid | — |
| Anthropic → OpenAI | Cost window throughput (req/s) | 4,603 | Valid | — |
| Anthropic → OpenAI | Cost window threads | 14 | Valid | — |
| Anthropic → OpenAI | Non-voluntary context switches per request | 0 | Valid | — |
| Anthropic → OpenAI | Cost window major page faults | 0 | Valid | — |
| Anthropic → OpenAI | Added time to first event p50 (µs) | 868 | Valid · n=100 gateway / 100 direct | — |
| Anthropic → OpenAI | Added time to first event p99 (µs) | 1,061 | Valid · n=100 gateway / 100 direct | — |
| Anthropic → OpenAI | Added inter-frame gap p50 (µs) | 4 | Valid · n=63 gateway / 63 direct | — |
| Anthropic → OpenAI | Added inter-frame gap p99 (µs) | 28 | Valid · n=63 gateway / 63 direct | — |
| Anthropic → OpenAI | Streams sustained (frames/s) | 29,329.1 | Valid | — |
| Anthropic → OpenAI | Streams sustained | 678 | Valid | — |
| Anthropic → OpenAI | Idle RSS (MiB) | 161 | Valid | — |
| Anthropic → OpenAI | Steady-state RSS (MiB) | 1,099.4 | Valid | — |
| Anthropic → OpenAI | Recovered RSS (MiB) | 1,073.3 | Valid | — |
| Anthropic → OpenAI | Peak RSS (MiB) | 1,300.5 | Valid | — |
| Anthropic → OpenAI | Peak RSS high-water mark (MiB) | 1,209.5 | Valid | — |
| Anthropic → OpenAI | Time to plateau (s) | — | Not measured | — |
| Anthropic → OpenAI | Memory growth (MiB/min) | 88.45 | Valid | — |
| Anthropic → OpenAI | Memory load duration (s) | 300 | Valid | — |
| Anthropic → OpenAI | Memory shape under load | 1 | Valid | — |
| Anthropic → OpenAI | Idle memory shape | — | Not measured | — |
| Anthropic → OpenAI | Idle memory static | 1 | Valid | — |
| Anthropic → OpenAI | Idle memory growth (MiB/min) | 0.93 | Valid | — |
| Anthropic → OpenAI | Throughput (req/s) | 1,531 | concurrency 1; n=9,192 | 1 ms |
| Anthropic → OpenAI | Throughput (req/s) | 4,343 | concurrency 4; n=26,062 | 5 ms |
| Anthropic → OpenAI | Throughput (req/s) | 4,579 | concurrency 8; n=27,481 | 10 ms |
| Anthropic → OpenAI | Throughput (req/s) | 4,623 | concurrency 16; n=27,743 | 50 ms |
| Anthropic → OpenAI | Throughput (req/s) | 4,623 | concurrency 16; n=27,743 | 100 ms |
| Anthropic → OpenAI | Throughput (req/s) | 4,623 | concurrency 16; n=27,743 | Unbounded |
| Gemini → OpenAI | Added latency p50 (µs) | 588 | Valid · n=9,465 gateway / 259,786 direct | — |
| Gemini → OpenAI | Added latency p99 (µs) | 972 | Valid · n=9,465 gateway / 259,786 direct | — |
| Gemini → OpenAI | Gateway p99 at concurrency 1 (µs) | 1,000 | Valid · n=9,465 | — |
| Gemini → OpenAI | Direct-to-mock p99 at concurrency 1 (µs) | 28 | Valid · n=259,786 | — |
| Gemini → OpenAI | CPU per request (µs) | 846.2 | Valid | — |
| Gemini → OpenAI | Requests per CPU-second | 1,181.7 | Valid | — |
| Gemini → OpenAI | Cost window concurrency | 8 | Valid | — |
| Gemini → OpenAI | Cost window core utilisation | 0.989 | Valid | — |
| Gemini → OpenAI | Cost window successful requests | 28,137 | Valid | — |
| Gemini → OpenAI | Cost window throughput (req/s) | 4,688 | Valid | — |
| Gemini → OpenAI | Cost window threads | 13 | Valid | — |
| Gemini → OpenAI | Non-voluntary context switches per request | 0.021 | Valid | — |
| Gemini → OpenAI | Cost window major page faults | 0 | Valid | — |
| Gemini → OpenAI | Added time to first event p50 (µs) | — | Untestable | — |
| Gemini → OpenAI | Added time to first event p99 (µs) | — | Untestable | — |
| Gemini → OpenAI | Added inter-frame gap p50 (µs) | — | Untestable | — |
| Gemini → OpenAI | Added inter-frame gap p99 (µs) | — | Untestable | — |
| Gemini → OpenAI | Streams sustained (frames/s) | — | Untestable | — |
| Gemini → OpenAI | Streams sustained | — | Untestable | — |
| Gemini → OpenAI | Idle RSS (MiB) | 162 | Valid | — |
| Gemini → OpenAI | Steady-state RSS (MiB) | 1,091.2 | Valid | — |
| Gemini → OpenAI | Recovered RSS (MiB) | 1,074.7 | Valid | — |
| Gemini → OpenAI | Peak RSS (MiB) | 1,249.8 | Valid | — |
| Gemini → OpenAI | Peak RSS high-water mark (MiB) | 1,190.7 | Valid | — |
| Gemini → OpenAI | Time to plateau (s) | — | Not measured | — |
| Gemini → OpenAI | Memory growth (MiB/min) | 81.08 | Valid | — |
| Gemini → OpenAI | Memory load duration (s) | 300 | Valid | — |
| Gemini → OpenAI | Memory shape under load | 1 | Valid | — |
| Gemini → OpenAI | Idle memory shape | — | Not measured | — |
| Gemini → OpenAI | Idle memory static | 1 | Valid | — |
| Gemini → OpenAI | Idle memory growth (MiB/min) | 0.263 | Valid | — |
| Gemini → OpenAI | Throughput (req/s) | 1,547 | concurrency 1; n=9,287 | 1 ms |
| Gemini → OpenAI | Throughput (req/s) | 4,450 | concurrency 4; n=26,704 | 5 ms |
| Gemini → OpenAI | Throughput (req/s) | 4,646 | concurrency 8; n=27,884 | 10 ms |
| Gemini → OpenAI | Throughput (req/s) | 4,724 | concurrency 16; n=28,353 | 50 ms |
| Gemini → OpenAI | Throughput (req/s) | 4,724 | concurrency 16; n=28,353 | 100 ms |
| Gemini → OpenAI | Throughput (req/s) | 4,724 | concurrency 16; n=28,353 | Unbounded |
| Cohere → OpenAI | Added latency p50 (µs) | 572 | Valid · n=9,654 gateway / 215,609 direct | — |
| Cohere → OpenAI | Added latency p99 (µs) | 912 | Valid · n=9,654 gateway / 215,609 direct | — |
| Cohere → OpenAI | Gateway p99 at concurrency 1 (µs) | 946 | Valid · n=9,654 | — |
| Cohere → OpenAI | Direct-to-mock p99 at concurrency 1 (µs) | 34 | Valid · n=215,609 | — |
| Cohere → OpenAI | CPU per request (µs) | 813.6 | Valid | — |
| Cohere → OpenAI | Requests per CPU-second | 1,229.1 | Valid | — |
| Cohere → OpenAI | Cost window concurrency | 8 | Valid | — |
| Cohere → OpenAI | Cost window core utilisation | 0.988 | Valid | — |
| Cohere → OpenAI | Cost window successful requests | 29,229 | Valid | — |
| Cohere → OpenAI | Cost window throughput (req/s) | 4,870 | Valid | — |
| Cohere → OpenAI | Cost window threads | 13 | Valid | — |
| Cohere → OpenAI | Non-voluntary context switches per request | 0.023 | Valid | — |
| Cohere → OpenAI | Cost window major page faults | 0 | Valid | — |
| Cohere → OpenAI | Added time to first event p50 (µs) | — | Untestable | — |
| Cohere → OpenAI | Added time to first event p99 (µs) | — | Untestable | — |
| Cohere → OpenAI | Added inter-frame gap p50 (µs) | — | Untestable | — |
| Cohere → OpenAI | Added inter-frame gap p99 (µs) | — | Untestable | — |
| Cohere → OpenAI | Streams sustained (frames/s) | — | Untestable | — |
| Cohere → OpenAI | Streams sustained | — | Untestable | — |
| Cohere → OpenAI | Idle RSS (MiB) | 158.4 | Valid | — |
| Cohere → OpenAI | Steady-state RSS (MiB) | 1,049.9 | Valid | — |
| Cohere → OpenAI | Recovered RSS (MiB) | 1,259.3 | Valid | — |
| Cohere → OpenAI | Peak RSS (MiB) | 1,259.3 | Valid | — |
| Cohere → OpenAI | Peak RSS high-water mark (MiB) | 1,138.4 | Valid | — |
| Cohere → OpenAI | Time to plateau (s) | — | Not measured | — |
| Cohere → OpenAI | Memory growth (MiB/min) | 78.13 | Valid | — |
| Cohere → OpenAI | Memory load duration (s) | 300 | Valid | — |
| Cohere → OpenAI | Memory shape under load | 1 | Valid | — |
| Cohere → OpenAI | Idle memory shape | — | Not measured | — |
| Cohere → OpenAI | Idle memory static | 1 | Valid | — |
| Cohere → OpenAI | Idle memory growth (MiB/min) | 0.933 | Valid | — |
| Cohere → OpenAI | Throughput (req/s) | 1,624 | concurrency 1; n=9,748 | 1 ms |
| Cohere → OpenAI | Throughput (req/s) | 4,853 | concurrency 8; n=29,122 | 5 ms |
| Cohere → OpenAI | Throughput (req/s) | 4,853 | concurrency 8; n=29,122 | 10 ms |
| Cohere → OpenAI | Throughput (req/s) | 4,898 | concurrency 16; n=29,394 | 50 ms |
| Cohere → OpenAI | Throughput (req/s) | 4,898 | concurrency 16; n=29,394 | 100 ms |
| Cohere → OpenAI | Throughput (req/s) | 4,898 | concurrency 16; n=29,394 | Unbounded |
| Bedrock → OpenAI | Added latency p50 (µs) | 632 | Valid · n=8,716 gateway / 262,374 direct | — |
| Bedrock → OpenAI | Added latency p99 (µs) | 1,148 | Valid · n=8,716 gateway / 262,374 direct | — |
| Bedrock → OpenAI | Gateway p99 at concurrency 1 (µs) | 1,176 | Valid · n=8,716 | — |
| Bedrock → OpenAI | Direct-to-mock p99 at concurrency 1 (µs) | 28 | Valid · n=262,374 | — |
| Bedrock → OpenAI | CPU per request (µs) | 909.3 | Valid | — |
| Bedrock → OpenAI | Requests per CPU-second | 1,099.8 | Valid | — |
| Bedrock → OpenAI | Cost window concurrency | 8 | Valid | — |
| Bedrock → OpenAI | Cost window core utilisation | 0.986 | Valid | — |
| Bedrock → OpenAI | Cost window successful requests | 26,109 | Valid | — |
| Bedrock → OpenAI | Cost window throughput (req/s) | 4,350 | Valid | — |
| Bedrock → OpenAI | Cost window threads | 13 | Valid | — |
| Bedrock → OpenAI | Non-voluntary context switches per request | 0.037 | Valid | — |
| Bedrock → OpenAI | Cost window major page faults | 0 | Valid | — |
| Bedrock → OpenAI | Added time to first event p50 (µs) | — | Untestable | — |
| Bedrock → OpenAI | Added time to first event p99 (µs) | — | Untestable | — |
| Bedrock → OpenAI | Added inter-frame gap p50 (µs) | — | Untestable | — |
| Bedrock → OpenAI | Added inter-frame gap p99 (µs) | — | Untestable | — |
| Bedrock → OpenAI | Streams sustained (frames/s) | — | Untestable | — |
| Bedrock → OpenAI | Streams sustained | — | Untestable | — |
| Bedrock → OpenAI | Idle RSS (MiB) | 160.4 | Valid | — |
| Bedrock → OpenAI | Steady-state RSS (MiB) | 1,042.7 | Valid | — |
| Bedrock → OpenAI | Recovered RSS (MiB) | 1,232 | Valid | — |
| Bedrock → OpenAI | Peak RSS (MiB) | 1,234.3 | Valid | — |
| Bedrock → OpenAI | Peak RSS high-water mark (MiB) | 1,146.6 | Valid | — |
| Bedrock → OpenAI | Time to plateau (s) | — | Not measured | — |
| Bedrock → OpenAI | Memory growth (MiB/min) | 73.77 | Valid | — |
| Bedrock → OpenAI | Memory load duration (s) | 300 | Valid | — |
| Bedrock → OpenAI | Memory shape under load | 1 | Valid | — |
| Bedrock → OpenAI | Idle memory shape | — | Not measured | — |
| Bedrock → OpenAI | Idle memory static | 1 | Valid | — |
| Bedrock → OpenAI | Idle memory growth (MiB/min) | 0.168 | Valid | — |
| Bedrock → OpenAI | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| Bedrock → OpenAI | Throughput (req/s) | 4,096 | concurrency 4; n=24,582 | 5 ms |
| Bedrock → OpenAI | Throughput (req/s) | 4,328 | concurrency 8; n=25,973 | 10 ms |
| Bedrock → OpenAI | Throughput (req/s) | 4,389 | concurrency 16; n=26,340 | 50 ms |
| Bedrock → OpenAI | Throughput (req/s) | 4,389 | concurrency 16; n=26,340 | 100 ms |
| Bedrock → OpenAI | Throughput (req/s) | 4,389 | concurrency 16; n=26,340 | Unbounded |
| OpenAI → OpenAI Responses | — | — | Not declared: recorded note 1 | — |
| OpenAI Responses → OpenAI Responses | Added latency p50 (µs) | 631 | Valid · n=8,878 gateway / 266,170 direct | — |
| OpenAI Responses → OpenAI Responses | Added latency p99 (µs) | 1,001 | Valid · n=8,878 gateway / 266,170 direct | — |
| OpenAI Responses → OpenAI Responses | Gateway p99 at concurrency 1 (µs) | 1,029 | Valid · n=8,878 | — |
| OpenAI Responses → OpenAI Responses | Direct-to-mock p99 at concurrency 1 (µs) | 28 | Valid · n=266,170 | — |
| OpenAI Responses → OpenAI Responses | CPU per request (µs) | 882.2 | Valid | — |
| OpenAI Responses → OpenAI Responses | Requests per CPU-second | 1,133.5 | Valid | — |
| OpenAI Responses → OpenAI Responses | Cost window concurrency | 8 | Valid | — |
| OpenAI Responses → OpenAI Responses | Cost window core utilisation | 0.99 | Valid | — |
| OpenAI Responses → OpenAI Responses | Cost window successful requests | 27,011 | Valid | — |
| OpenAI Responses → OpenAI Responses | Cost window throughput (req/s) | 4,501 | Valid | — |
| OpenAI Responses → OpenAI Responses | Cost window threads | 15 | Valid | — |
| OpenAI Responses → OpenAI Responses | Non-voluntary context switches per request | 0.02 | Valid | — |
| OpenAI Responses → OpenAI Responses | Cost window major page faults | 0 | Valid | — |
| OpenAI Responses → OpenAI Responses | Added time to first event p50 (µs) | — | Untestable | — |
| OpenAI Responses → OpenAI Responses | Added time to first event p99 (µs) | — | Untestable | — |
| OpenAI Responses → OpenAI Responses | Added inter-frame gap p50 (µs) | — | Untestable | — |
| OpenAI Responses → OpenAI Responses | Added inter-frame gap p99 (µs) | — | Untestable | — |
| OpenAI Responses → OpenAI Responses | Streams sustained (frames/s) | — | Untestable | — |
| OpenAI Responses → OpenAI Responses | Streams sustained | — | Untestable | — |
| OpenAI Responses → OpenAI Responses | Idle RSS (MiB) | 160.6 | Valid | — |
| OpenAI Responses → OpenAI Responses | Steady-state RSS (MiB) | 1,093.7 | Valid | — |
| OpenAI Responses → OpenAI Responses | Recovered RSS (MiB) | 1,288 | Valid | — |
| OpenAI Responses → OpenAI Responses | Peak RSS (MiB) | 1,288.1 | Valid | — |
| OpenAI Responses → OpenAI Responses | Peak RSS high-water mark (MiB) | 1,178.1 | Valid | — |
| OpenAI Responses → OpenAI Responses | Time to plateau (s) | — | Not measured | — |
| OpenAI Responses → OpenAI Responses | Memory growth (MiB/min) | 64.42 | Valid | — |
| OpenAI Responses → OpenAI Responses | Memory load duration (s) | 300 | Valid | — |
| OpenAI Responses → OpenAI Responses | Memory shape under load | 1 | Valid | — |
| OpenAI Responses → OpenAI Responses | Idle memory shape | — | Not measured | — |
| OpenAI Responses → OpenAI Responses | Idle memory static | 1 | Valid | — |
| OpenAI Responses → OpenAI Responses | Idle memory growth (MiB/min) | 0.224 | Valid | — |
| OpenAI Responses → OpenAI Responses | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| OpenAI Responses → OpenAI Responses | Throughput (req/s) | 4,320 | concurrency 4; n=25,924 | 5 ms |
| OpenAI Responses → OpenAI Responses | Throughput (req/s) | 4,443 | concurrency 8; n=26,666 | 10 ms |
| OpenAI Responses → OpenAI Responses | Throughput (req/s) | 4,471 | concurrency 16; n=26,831 | 50 ms |
| OpenAI Responses → OpenAI Responses | Throughput (req/s) | 4,471 | concurrency 16; n=26,831 | 100 ms |
| OpenAI Responses → OpenAI Responses | Throughput (req/s) | 4,471 | concurrency 16; n=26,831 | Unbounded |
| Anthropic → OpenAI Responses | Added latency p50 (µs) | 641 | Valid · n=8,756 gateway / 262,163 direct | — |
| Anthropic → OpenAI Responses | Added latency p99 (µs) | 1,011 | Valid · n=8,756 gateway / 262,163 direct | — |
| Anthropic → OpenAI Responses | Gateway p99 at concurrency 1 (µs) | 1,038 | Valid · n=8,756 | — |
| Anthropic → OpenAI Responses | Direct-to-mock p99 at concurrency 1 (µs) | 27 | Valid · n=262,163 | — |
| Anthropic → OpenAI Responses | CPU per request (µs) | 886.3 | Valid | — |
| Anthropic → OpenAI Responses | Requests per CPU-second | 1,128.2 | Valid | — |
| Anthropic → OpenAI Responses | Cost window concurrency | 8 | Valid | — |
| Anthropic → OpenAI Responses | Cost window core utilisation | 0.989 | Valid | — |
| Anthropic → OpenAI Responses | Cost window successful requests | 26,852 | Valid | — |
| Anthropic → OpenAI Responses | Cost window throughput (req/s) | 4,474 | Valid | — |
| Anthropic → OpenAI Responses | Cost window threads | 13 | Valid | — |
| Anthropic → OpenAI Responses | Non-voluntary context switches per request | 0.023 | Valid | — |
| Anthropic → OpenAI Responses | Cost window major page faults | 0 | Valid | — |
| Anthropic → OpenAI Responses | Added time to first event p50 (µs) | — | Untestable | — |
| Anthropic → OpenAI Responses | Added time to first event p99 (µs) | — | Untestable | — |
| Anthropic → OpenAI Responses | Added inter-frame gap p50 (µs) | — | Untestable | — |
| Anthropic → OpenAI Responses | Added inter-frame gap p99 (µs) | — | Untestable | — |
| Anthropic → OpenAI Responses | Streams sustained (frames/s) | — | Untestable | — |
| Anthropic → OpenAI Responses | Streams sustained | — | Untestable | — |
| Anthropic → OpenAI Responses | Idle RSS (MiB) | 164.2 | Valid | — |
| Anthropic → OpenAI Responses | Steady-state RSS (MiB) | 1,105.5 | Valid | — |
| Anthropic → OpenAI Responses | Recovered RSS (MiB) | 1,137.9 | Valid | — |
| Anthropic → OpenAI Responses | Peak RSS (MiB) | 1,237.3 | Valid | — |
| Anthropic → OpenAI Responses | Peak RSS high-water mark (MiB) | 1,209.7 | Valid | — |
| Anthropic → OpenAI Responses | Time to plateau (s) | — | Not measured | — |
| Anthropic → OpenAI Responses | Memory growth (MiB/min) | 83.21 | Valid | — |
| Anthropic → OpenAI Responses | Memory load duration (s) | 300 | Valid | — |
| Anthropic → OpenAI Responses | Memory shape under load | 1 | Valid | — |
| Anthropic → OpenAI Responses | Idle memory shape | — | Not measured | — |
| Anthropic → OpenAI Responses | Idle memory static | 1 | Valid | — |
| Anthropic → OpenAI Responses | Idle memory growth (MiB/min) | 0.97 | Valid | — |
| Anthropic → OpenAI Responses | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| Anthropic → OpenAI Responses | Throughput (req/s) | 4,161 | concurrency 4; n=24,972 | 5 ms |
| Anthropic → OpenAI Responses | Throughput (req/s) | 4,330 | concurrency 8; n=25,985 | 10 ms |
| Anthropic → OpenAI Responses | Throughput (req/s) | 4,391 | concurrency 16; n=26,357 | 50 ms |
| Anthropic → OpenAI Responses | Throughput (req/s) | 4,391 | concurrency 16; n=26,357 | 100 ms |
| Anthropic → OpenAI Responses | Throughput (req/s) | 4,391 | concurrency 16; n=26,357 | Unbounded |
| Gemini → OpenAI Responses | Added latency p50 (µs) | 653 | Valid · n=8,613 gateway / 269,135 direct | — |
| Gemini → OpenAI Responses | Added latency p99 (µs) | 1,008 | Valid · n=8,613 gateway / 269,135 direct | — |
| Gemini → OpenAI Responses | Gateway p99 at concurrency 1 (µs) | 1,035 | Valid · n=8,613 | — |
| Gemini → OpenAI Responses | Direct-to-mock p99 at concurrency 1 (µs) | 27 | Valid · n=269,135 | — |
| Gemini → OpenAI Responses | CPU per request (µs) | 901.3 | Valid | — |
| Gemini → OpenAI Responses | Requests per CPU-second | 1,109.5 | Valid | — |
| Gemini → OpenAI Responses | Cost window concurrency | 8 | Valid | — |
| Gemini → OpenAI Responses | Cost window core utilisation | 0.989 | Valid | — |
| Gemini → OpenAI Responses | Cost window successful requests | 26,417 | Valid | — |
| Gemini → OpenAI Responses | Cost window throughput (req/s) | 4,402 | Valid | — |
| Gemini → OpenAI Responses | Cost window threads | 14 | Valid | — |
| Gemini → OpenAI Responses | Non-voluntary context switches per request | 0 | Valid | — |
| Gemini → OpenAI Responses | Cost window major page faults | 0 | Valid | — |
| Gemini → OpenAI Responses | Added time to first event p50 (µs) | — | Untestable | — |
| Gemini → OpenAI Responses | Added time to first event p99 (µs) | — | Untestable | — |
| Gemini → OpenAI Responses | Added inter-frame gap p50 (µs) | — | Untestable | — |
| Gemini → OpenAI Responses | Added inter-frame gap p99 (µs) | — | Untestable | — |
| Gemini → OpenAI Responses | Streams sustained (frames/s) | — | Untestable | — |
| Gemini → OpenAI Responses | Streams sustained | — | Untestable | — |
| Gemini → OpenAI Responses | Idle RSS (MiB) | 159.7 | Valid | — |
| Gemini → OpenAI Responses | Steady-state RSS (MiB) | 1,083.4 | Valid | — |
| Gemini → OpenAI Responses | Recovered RSS (MiB) | 1,240 | Valid | — |
| Gemini → OpenAI Responses | Peak RSS (MiB) | 1,240 | Valid | — |
| Gemini → OpenAI Responses | Peak RSS high-water mark (MiB) | 1,183.2 | Valid | — |
| Gemini → OpenAI Responses | Time to plateau (s) | — | Not measured | — |
| Gemini → OpenAI Responses | Memory growth (MiB/min) | 66.32 | Valid | — |
| Gemini → OpenAI Responses | Memory load duration (s) | 300 | Valid | — |
| Gemini → OpenAI Responses | Memory shape under load | 1 | Valid | — |
| Gemini → OpenAI Responses | Idle memory shape | — | Not measured | — |
| Gemini → OpenAI Responses | Idle memory static | 1 | Valid | — |
| Gemini → OpenAI Responses | Idle memory growth (MiB/min) | 0.569 | Valid | — |
| Gemini → OpenAI Responses | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| Gemini → OpenAI Responses | Throughput (req/s) | 4,182 | concurrency 4; n=25,101 | 5 ms |
| Gemini → OpenAI Responses | Throughput (req/s) | 4,323 | concurrency 8; n=25,944 | 10 ms |
| Gemini → OpenAI Responses | Throughput (req/s) | 4,375 | concurrency 16; n=26,254 | 50 ms |
| Gemini → OpenAI Responses | Throughput (req/s) | 4,375 | concurrency 16; n=26,254 | 100 ms |
| Gemini → OpenAI Responses | Throughput (req/s) | 4,375 | concurrency 16; n=26,254 | Unbounded |
| Cohere → OpenAI Responses | — | — | Not declared: recorded note 1 | — |
| Bedrock → OpenAI Responses | Added latency p50 (µs) | 611 | Valid · n=9,156 gateway / 264,307 direct | — |
| Bedrock → OpenAI Responses | Added latency p99 (µs) | 958 | Valid · n=9,156 gateway / 264,307 direct | — |
| Bedrock → OpenAI Responses | Gateway p99 at concurrency 1 (µs) | 985 | Valid · n=9,156 | — |
| Bedrock → OpenAI Responses | Direct-to-mock p99 at concurrency 1 (µs) | 27 | Valid · n=264,307 | — |
| Bedrock → OpenAI Responses | CPU per request (µs) | 851.8 | Valid | — |
| Bedrock → OpenAI Responses | Requests per CPU-second | 1,174 | Valid | — |
| Bedrock → OpenAI Responses | Cost window concurrency | 8 | Valid | — |
| Bedrock → OpenAI Responses | Cost window core utilisation | 0.993 | Valid | — |
| Bedrock → OpenAI Responses | Cost window successful requests | 28,082 | Valid | — |
| Bedrock → OpenAI Responses | Cost window throughput (req/s) | 4,679 | Valid | — |
| Bedrock → OpenAI Responses | Cost window threads | 14 | Valid | — |
| Bedrock → OpenAI Responses | Non-voluntary context switches per request | 0.044 | Valid | — |
| Bedrock → OpenAI Responses | Cost window major page faults | 0 | Valid | — |
| Bedrock → OpenAI Responses | Added time to first event p50 (µs) | — | Untestable | — |
| Bedrock → OpenAI Responses | Added time to first event p99 (µs) | — | Untestable | — |
| Bedrock → OpenAI Responses | Added inter-frame gap p50 (µs) | — | Untestable | — |
| Bedrock → OpenAI Responses | Added inter-frame gap p99 (µs) | — | Untestable | — |
| Bedrock → OpenAI Responses | Streams sustained (frames/s) | — | Untestable | — |
| Bedrock → OpenAI Responses | Streams sustained | — | Untestable | — |
| Bedrock → OpenAI Responses | Idle RSS (MiB) | 161.6 | Valid | — |
| Bedrock → OpenAI Responses | Steady-state RSS (MiB) | 1,083.2 | Valid | — |
| Bedrock → OpenAI Responses | Recovered RSS (MiB) | 1,295.6 | Valid | — |
| Bedrock → OpenAI Responses | Peak RSS (MiB) | 1,295.6 | Valid | — |
| Bedrock → OpenAI Responses | Peak RSS high-water mark (MiB) | 1,184 | Valid | — |
| Bedrock → OpenAI Responses | Time to plateau (s) | — | Not measured | — |
| Bedrock → OpenAI Responses | Memory growth (MiB/min) | 85.13 | Valid | — |
| Bedrock → OpenAI Responses | Memory load duration (s) | 300 | Valid | — |
| Bedrock → OpenAI Responses | Memory shape under load | 1 | Valid | — |
| Bedrock → OpenAI Responses | Idle memory shape | — | Not measured | — |
| Bedrock → OpenAI Responses | Idle memory static | 1 | Valid | — |
| Bedrock → OpenAI Responses | Idle memory growth (MiB/min) | 0.526 | Valid | — |
| Bedrock → OpenAI Responses | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| Bedrock → OpenAI Responses | Throughput (req/s) | 4,197 | concurrency 4; n=25,188 | 5 ms |
| Bedrock → OpenAI Responses | Throughput (req/s) | 4,442 | concurrency 8; n=26,658 | 10 ms |
| Bedrock → OpenAI Responses | Throughput (req/s) | 4,547 | concurrency 32; n=27,285 | 50 ms |
| Bedrock → OpenAI Responses | Throughput (req/s) | 4,547 | concurrency 32; n=27,285 | 100 ms |
| Bedrock → OpenAI Responses | Throughput (req/s) | 4,547 | concurrency 32; n=27,285 | Unbounded |
| OpenAI → Anthropic | Added latency p50 (µs) | 673 | Valid · n=8,297 gateway / 262,017 direct | — |
| OpenAI → Anthropic | Added latency p99 (µs) | 1,094 | Valid · n=8,297 gateway / 262,017 direct | — |
| OpenAI → Anthropic | Gateway p99 at concurrency 1 (µs) | 1,122 | Valid · n=8,297 | — |
| OpenAI → Anthropic | Direct-to-mock p99 at concurrency 1 (µs) | 28 | Valid · n=262,017 | — |
| OpenAI → Anthropic | CPU per request (µs) | 931.3 | Valid | — |
| OpenAI → Anthropic | Requests per CPU-second | 1,073.7 | Valid | — |
| OpenAI → Anthropic | Cost window concurrency | 8 | Valid | — |
| OpenAI → Anthropic | Cost window core utilisation | 0.991 | Valid | — |
| OpenAI → Anthropic | Cost window successful requests | 25,608 | Valid | — |
| OpenAI → Anthropic | Cost window throughput (req/s) | 4,267 | Valid | — |
| OpenAI → Anthropic | Cost window threads | 13 | Valid | — |
| OpenAI → Anthropic | Non-voluntary context switches per request | 0.022 | Valid | — |
| OpenAI → Anthropic | Cost window major page faults | 0 | Valid | — |
| OpenAI → Anthropic | Added time to first event p50 (µs) | 857 | Valid · n=100 gateway / 100 direct | — |
| OpenAI → Anthropic | Added time to first event p99 (µs) | 1,039 | Valid · n=100 gateway / 100 direct | — |
| OpenAI → Anthropic | Added inter-frame gap p50 (µs) | 7 | Valid · n=63 gateway / 63 direct | — |
| OpenAI → Anthropic | Added inter-frame gap p99 (µs) | 25 | Valid · n=63 gateway / 63 direct | — |
| OpenAI → Anthropic | Streams sustained (frames/s) | 26,538 | Valid | — |
| OpenAI → Anthropic | Streams sustained | 624 | Valid | — |
| OpenAI → Anthropic | Idle RSS (MiB) | 162.3 | Valid | — |
| OpenAI → Anthropic | Steady-state RSS (MiB) | 1,067.1 | Valid | — |
| OpenAI → Anthropic | Recovered RSS (MiB) | 1,319.3 | Valid | — |
| OpenAI → Anthropic | Peak RSS (MiB) | 1,319.3 | Valid | — |
| OpenAI → Anthropic | Peak RSS high-water mark (MiB) | 1,199.9 | Valid | — |
| OpenAI → Anthropic | Time to plateau (s) | — | Not measured | — |
| OpenAI → Anthropic | Memory growth (MiB/min) | 63.29 | Valid | — |
| OpenAI → Anthropic | Memory load duration (s) | 300 | Valid | — |
| OpenAI → Anthropic | Memory shape under load | 1 | Valid | — |
| OpenAI → Anthropic | Idle memory shape | — | Not measured | — |
| OpenAI → Anthropic | Idle memory static | 1 | Valid | — |
| OpenAI → Anthropic | Idle memory growth (MiB/min) | 0.634 | Valid | — |
| OpenAI → Anthropic | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| OpenAI → Anthropic | Throughput (req/s) | 4,001 | concurrency 4; n=24,014 | 5 ms |
| OpenAI → Anthropic | Throughput (req/s) | 4,117 | concurrency 8; n=24,710 | 10 ms |
| OpenAI → Anthropic | Throughput (req/s) | 4,177 | concurrency 32; n=25,067 | 50 ms |
| OpenAI → Anthropic | Throughput (req/s) | 4,177 | concurrency 32; n=25,067 | 100 ms |
| OpenAI → Anthropic | Throughput (req/s) | 4,177 | concurrency 32; n=25,067 | Unbounded |
| OpenAI Responses → Anthropic | Added latency p50 (µs) | 770 | Valid · n=7,332 gateway / 263,386 direct | — |
| OpenAI Responses → Anthropic | Added latency p99 (µs) | 1,259 | Valid · n=7,332 gateway / 263,386 direct | — |
| OpenAI Responses → Anthropic | Gateway p99 at concurrency 1 (µs) | 1,287 | Valid · n=7,332 | — |
| OpenAI Responses → Anthropic | Direct-to-mock p99 at concurrency 1 (µs) | 28 | Valid · n=263,386 | — |
| OpenAI Responses → Anthropic | CPU per request (µs) | 1,057.1 | Valid | — |
| OpenAI Responses → Anthropic | Requests per CPU-second | 946 | Valid | — |
| OpenAI Responses → Anthropic | Cost window concurrency | 8 | Valid | — |
| OpenAI Responses → Anthropic | Cost window core utilisation | 0.986 | Valid | — |
| OpenAI Responses → Anthropic | Cost window successful requests | 22,448 | Valid | — |
| OpenAI Responses → Anthropic | Cost window throughput (req/s) | 3,740 | Valid | — |
| OpenAI Responses → Anthropic | Cost window threads | 13 | Valid | — |
| OpenAI Responses → Anthropic | Non-voluntary context switches per request | 0 | Valid | — |
| OpenAI Responses → Anthropic | Cost window major page faults | 0 | Valid | — |
| OpenAI Responses → Anthropic | Added time to first event p50 (µs) | — | Untestable | — |
| OpenAI Responses → Anthropic | Added time to first event p99 (µs) | — | Untestable | — |
| OpenAI Responses → Anthropic | Added inter-frame gap p50 (µs) | — | Untestable | — |
| OpenAI Responses → Anthropic | Added inter-frame gap p99 (µs) | — | Untestable | — |
| OpenAI Responses → Anthropic | Streams sustained (frames/s) | — | Untestable | — |
| OpenAI Responses → Anthropic | Streams sustained | — | Untestable | — |
| OpenAI Responses → Anthropic | Idle RSS (MiB) | 159.1 | Valid | — |
| OpenAI Responses → Anthropic | Steady-state RSS (MiB) | 984.9 | Valid | — |
| OpenAI Responses → Anthropic | Recovered RSS (MiB) | 1,051 | Valid | — |
| OpenAI Responses → Anthropic | Peak RSS (MiB) | 1,220.6 | Valid | — |
| OpenAI Responses → Anthropic | Peak RSS high-water mark (MiB) | 1,174.6 | Valid | — |
| OpenAI Responses → Anthropic | Time to plateau (s) | — | Not measured | — |
| OpenAI Responses → Anthropic | Memory growth (MiB/min) | 168.5 | Valid | — |
| OpenAI Responses → Anthropic | Memory load duration (s) | 300 | Valid | — |
| OpenAI Responses → Anthropic | Memory shape under load | 1 | Valid | — |
| OpenAI Responses → Anthropic | Idle memory shape | — | Not measured | — |
| OpenAI Responses → Anthropic | Idle memory static | 1 | Valid | — |
| OpenAI Responses → Anthropic | Idle memory growth (MiB/min) | 0.47 | Valid | — |
| OpenAI Responses → Anthropic | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| OpenAI Responses → Anthropic | Throughput (req/s) | 3,458 | concurrency 4; n=20,760 | 5 ms |
| OpenAI Responses → Anthropic | Throughput (req/s) | 3,703 | concurrency 8; n=22,224 | 10 ms |
| OpenAI Responses → Anthropic | Throughput (req/s) | 3,703 | concurrency 8; n=22,224 | 50 ms |
| OpenAI Responses → Anthropic | Throughput (req/s) | 3,703 | concurrency 8; n=22,224 | 100 ms |
| OpenAI Responses → Anthropic | Throughput (req/s) | 3,703 | concurrency 8; n=22,224 | Unbounded |
| Anthropic → Anthropic | Added latency p50 (µs) | 678 | Valid · n=8,283 gateway / 258,757 direct | — |
| Anthropic → Anthropic | Added latency p99 (µs) | 1,101 | Valid · n=8,283 gateway / 258,757 direct | — |
| Anthropic → Anthropic | Gateway p99 at concurrency 1 (µs) | 1,129 | Valid · n=8,283 | — |
| Anthropic → Anthropic | Direct-to-mock p99 at concurrency 1 (µs) | 28 | Valid · n=258,757 | — |
| Anthropic → Anthropic | CPU per request (µs) | 1,001.9 | Valid | — |
| Anthropic → Anthropic | Requests per CPU-second | 998.1 | Valid | — |
| Anthropic → Anthropic | Cost window concurrency | 8 | Valid | — |
| Anthropic → Anthropic | Cost window core utilisation | 0.995 | Valid | — |
| Anthropic → Anthropic | Cost window successful requests | 23,914 | Valid | — |
| Anthropic → Anthropic | Cost window throughput (req/s) | 3,985 | Valid | — |
| Anthropic → Anthropic | Cost window threads | 13 | Valid | — |
| Anthropic → Anthropic | Non-voluntary context switches per request | 0.037 | Valid | — |
| Anthropic → Anthropic | Cost window major page faults | 0 | Valid | — |
| Anthropic → Anthropic | Added time to first event p50 (µs) | 826 | Valid · n=100 gateway / 100 direct | — |
| Anthropic → Anthropic | Added time to first event p99 (µs) | 924 | Valid · n=100 gateway / 100 direct | — |
| Anthropic → Anthropic | Added inter-frame gap p50 (µs) | — | Below resolution · n=63 gateway / 63 direct | — |
| Anthropic → Anthropic | Added inter-frame gap p99 (µs) | — | Below resolution · n=63 gateway / 63 direct | — |
| Anthropic → Anthropic | Streams sustained (frames/s) | 28,037.5 | Valid | — |
| Anthropic → Anthropic | Streams sustained | 683 | Valid | — |
| Anthropic → Anthropic | Idle RSS (MiB) | 161.9 | Valid | — |
| Anthropic → Anthropic | Steady-state RSS (MiB) | 1,059.2 | Valid | — |
| Anthropic → Anthropic | Recovered RSS (MiB) | 1,112.5 | Valid | — |
| Anthropic → Anthropic | Peak RSS (MiB) | 1,274.2 | Valid | — |
| Anthropic → Anthropic | Peak RSS high-water mark (MiB) | 1,215.4 | Valid | — |
| Anthropic → Anthropic | Time to plateau (s) | — | Not measured | — |
| Anthropic → Anthropic | Memory growth (MiB/min) | 80.73 | Valid | — |
| Anthropic → Anthropic | Memory load duration (s) | 300 | Valid | — |
| Anthropic → Anthropic | Memory shape under load | 1 | Valid | — |
| Anthropic → Anthropic | Idle memory shape | — | Not measured | — |
| Anthropic → Anthropic | Idle memory static | 1 | Valid | — |
| Anthropic → Anthropic | Idle memory growth (MiB/min) | 0.715 | Valid | — |
| Anthropic → Anthropic | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| Anthropic → Anthropic | Throughput (req/s) | 3,861 | concurrency 4; n=23,174 | 5 ms |
| Anthropic → Anthropic | Throughput (req/s) | 3,988 | concurrency 8; n=23,931 | 10 ms |
| Anthropic → Anthropic | Throughput (req/s) | 4,035 | concurrency 16; n=24,216 | 50 ms |
| Anthropic → Anthropic | Throughput (req/s) | 4,035 | concurrency 16; n=24,216 | 100 ms |
| Anthropic → Anthropic | Throughput (req/s) | 4,035 | concurrency 16; n=24,216 | Unbounded |
| Gemini → Anthropic | Added latency p50 (µs) | 693 | Valid · n=8,044 gateway / 259,231 direct | — |
| Gemini → Anthropic | Added latency p99 (µs) | 1,153 | Valid · n=8,044 gateway / 259,231 direct | — |
| Gemini → Anthropic | Gateway p99 at concurrency 1 (µs) | 1,181 | Valid · n=8,044 | — |
| Gemini → Anthropic | Direct-to-mock p99 at concurrency 1 (µs) | 28 | Valid · n=259,231 | — |
| Gemini → Anthropic | CPU per request (µs) | 999 | Valid | — |
| Gemini → Anthropic | Requests per CPU-second | 1,001 | Valid | — |
| Gemini → Anthropic | Cost window concurrency | 8 | Valid | — |
| Gemini → Anthropic | Cost window core utilisation | 0.994 | Valid | — |
| Gemini → Anthropic | Cost window successful requests | 23,954 | Valid | — |
| Gemini → Anthropic | Cost window throughput (req/s) | 3,991 | Valid | — |
| Gemini → Anthropic | Cost window threads | 16 | Valid | — |
| Gemini → Anthropic | Non-voluntary context switches per request | 0 | Valid | — |
| Gemini → Anthropic | Cost window major page faults | 0 | Valid | — |
| Gemini → Anthropic | Added time to first event p50 (µs) | — | Untestable | — |
| Gemini → Anthropic | Added time to first event p99 (µs) | — | Untestable | — |
| Gemini → Anthropic | Added inter-frame gap p50 (µs) | — | Untestable | — |
| Gemini → Anthropic | Added inter-frame gap p99 (µs) | — | Untestable | — |
| Gemini → Anthropic | Streams sustained (frames/s) | — | Untestable | — |
| Gemini → Anthropic | Streams sustained | — | Untestable | — |
| Gemini → Anthropic | Idle RSS (MiB) | 161.1 | Valid | — |
| Gemini → Anthropic | Steady-state RSS (MiB) | 1,047.8 | Valid | — |
| Gemini → Anthropic | Recovered RSS (MiB) | 1,109.1 | Valid | — |
| Gemini → Anthropic | Peak RSS (MiB) | 1,249.8 | Valid | — |
| Gemini → Anthropic | Peak RSS high-water mark (MiB) | 1,215.3 | Valid | — |
| Gemini → Anthropic | Time to plateau (s) | — | Not measured | — |
| Gemini → Anthropic | Memory growth (MiB/min) | 96.98 | Valid | — |
| Gemini → Anthropic | Memory load duration (s) | 300 | Valid | — |
| Gemini → Anthropic | Memory shape under load | 1 | Valid | — |
| Gemini → Anthropic | Idle memory shape | — | Not measured | — |
| Gemini → Anthropic | Idle memory static | 1 | Valid | — |
| Gemini → Anthropic | Idle memory growth (MiB/min) | 0.212 | Valid | — |
| Gemini → Anthropic | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| Gemini → Anthropic | Throughput (req/s) | 3,872 | concurrency 4; n=23,239 | 5 ms |
| Gemini → Anthropic | Throughput (req/s) | 3,995 | concurrency 8; n=23,976 | 10 ms |
| Gemini → Anthropic | Throughput (req/s) | 4,040 | concurrency 16; n=24,249 | 50 ms |
| Gemini → Anthropic | Throughput (req/s) | 4,040 | concurrency 16; n=24,249 | 100 ms |
| Gemini → Anthropic | Throughput (req/s) | 4,040 | concurrency 16; n=24,249 | Unbounded |
| Cohere → Anthropic | Added latency p50 (µs) | 685 | Valid · n=8,100 gateway / 245,712 direct | — |
| Cohere → Anthropic | Added latency p99 (µs) | 1,201 | Valid · n=8,100 gateway / 245,712 direct | — |
| Cohere → Anthropic | Gateway p99 at concurrency 1 (µs) | 1,230 | Valid · n=8,100 | — |
| Cohere → Anthropic | Direct-to-mock p99 at concurrency 1 (µs) | 29 | Valid · n=245,712 | — |
| Cohere → Anthropic | CPU per request (µs) | 946.2 | Valid | — |
| Cohere → Anthropic | Requests per CPU-second | 1,056.9 | Valid | — |
| Cohere → Anthropic | Cost window concurrency | 8 | Valid | — |
| Cohere → Anthropic | Cost window core utilisation | 0.991 | Valid | — |
| Cohere → Anthropic | Cost window successful requests | 25,217 | Valid | — |
| Cohere → Anthropic | Cost window throughput (req/s) | 4,202 | Valid | — |
| Cohere → Anthropic | Cost window threads | 13 | Valid | — |
| Cohere → Anthropic | Non-voluntary context switches per request | 0.02 | Valid | — |
| Cohere → Anthropic | Cost window major page faults | 0 | Valid | — |
| Cohere → Anthropic | Added time to first event p50 (µs) | — | Untestable | — |
| Cohere → Anthropic | Added time to first event p99 (µs) | — | Untestable | — |
| Cohere → Anthropic | Added inter-frame gap p50 (µs) | — | Untestable | — |
| Cohere → Anthropic | Added inter-frame gap p99 (µs) | — | Untestable | — |
| Cohere → Anthropic | Streams sustained (frames/s) | — | Untestable | — |
| Cohere → Anthropic | Streams sustained | — | Untestable | — |
| Cohere → Anthropic | Idle RSS (MiB) | 163 | Valid | — |
| Cohere → Anthropic | Steady-state RSS (MiB) | 1,019.6 | Valid | — |
| Cohere → Anthropic | Recovered RSS (MiB) | 961.4 | Valid | — |
| Cohere → Anthropic | Peak RSS (MiB) | 1,261.3 | Valid | — |
| Cohere → Anthropic | Peak RSS high-water mark (MiB) | 1,159 | Valid | — |
| Cohere → Anthropic | Time to plateau (s) | — | Not measured | — |
| Cohere → Anthropic | Memory growth (MiB/min) | 69.15 | Valid | — |
| Cohere → Anthropic | Memory load duration (s) | 300 | Valid | — |
| Cohere → Anthropic | Memory shape under load | 1 | Valid | — |
| Cohere → Anthropic | Idle memory shape | — | Not measured | — |
| Cohere → Anthropic | Idle memory static | 1 | Valid | — |
| Cohere → Anthropic | Idle memory growth (MiB/min) | 0.268 | Valid | — |
| Cohere → Anthropic | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| Cohere → Anthropic | Throughput (req/s) | 3,987 | concurrency 4; n=23,927 | 5 ms |
| Cohere → Anthropic | Throughput (req/s) | 4,149 | concurrency 8; n=24,902 | 10 ms |
| Cohere → Anthropic | Throughput (req/s) | 4,224 | concurrency 16; n=25,349 | 50 ms |
| Cohere → Anthropic | Throughput (req/s) | 4,224 | concurrency 16; n=25,349 | 100 ms |
| Cohere → Anthropic | Throughput (req/s) | 4,224 | concurrency 16; n=25,349 | Unbounded |
| Bedrock → Anthropic | Added latency p50 (µs) | 620 | Valid · n=8,987 gateway / 265,983 direct | — |
| Bedrock → Anthropic | Added latency p99 (µs) | 1,030 | Valid · n=8,987 gateway / 265,983 direct | — |
| Bedrock → Anthropic | Gateway p99 at concurrency 1 (µs) | 1,057 | Valid · n=8,987 | — |
| Bedrock → Anthropic | Direct-to-mock p99 at concurrency 1 (µs) | 27 | Valid · n=265,983 | — |
| Bedrock → Anthropic | CPU per request (µs) | 869.5 | Valid | — |
| Bedrock → Anthropic | Requests per CPU-second | 1,150.1 | Valid | — |
| Bedrock → Anthropic | Cost window concurrency | 8 | Valid | — |
| Bedrock → Anthropic | Cost window core utilisation | 0.991 | Valid | — |
| Bedrock → Anthropic | Cost window successful requests | 27,453 | Valid | — |
| Bedrock → Anthropic | Cost window throughput (req/s) | 4,574 | Valid | — |
| Bedrock → Anthropic | Cost window threads | 13 | Valid | — |
| Bedrock → Anthropic | Non-voluntary context switches per request | 0 | Valid | — |
| Bedrock → Anthropic | Cost window major page faults | 0 | Valid | — |
| Bedrock → Anthropic | Added time to first event p50 (µs) | — | Untestable | — |
| Bedrock → Anthropic | Added time to first event p99 (µs) | — | Untestable | — |
| Bedrock → Anthropic | Added inter-frame gap p50 (µs) | — | Untestable | — |
| Bedrock → Anthropic | Added inter-frame gap p99 (µs) | — | Untestable | — |
| Bedrock → Anthropic | Streams sustained (frames/s) | — | Untestable | — |
| Bedrock → Anthropic | Streams sustained | — | Untestable | — |
| Bedrock → Anthropic | Idle RSS (MiB) | 160.3 | Valid | — |
| Bedrock → Anthropic | Steady-state RSS (MiB) | 1,059.3 | Valid | — |
| Bedrock → Anthropic | Recovered RSS (MiB) | 1,284 | Valid | — |
| Bedrock → Anthropic | Peak RSS (MiB) | 1,284 | Valid | — |
| Bedrock → Anthropic | Peak RSS high-water mark (MiB) | 1,156 | Valid | — |
| Bedrock → Anthropic | Time to plateau (s) | — | Not measured | — |
| Bedrock → Anthropic | Memory growth (MiB/min) | 44.08 | Valid | — |
| Bedrock → Anthropic | Memory load duration (s) | 300 | Valid | — |
| Bedrock → Anthropic | Memory shape under load | 1 | Valid | — |
| Bedrock → Anthropic | Idle memory shape | — | Not measured | — |
| Bedrock → Anthropic | Idle memory static | 1 | Valid | — |
| Bedrock → Anthropic | Idle memory growth (MiB/min) | 0.56 | Valid | — |
| Bedrock → Anthropic | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| Bedrock → Anthropic | Throughput (req/s) | 4,524 | concurrency 8; n=27,150 | 5 ms |
| Bedrock → Anthropic | Throughput (req/s) | 4,524 | concurrency 8; n=27,150 | 10 ms |
| Bedrock → Anthropic | Throughput (req/s) | 4,633 | concurrency 16; n=27,807 | 50 ms |
| Bedrock → Anthropic | Throughput (req/s) | 4,633 | concurrency 16; n=27,807 | 100 ms |
| Bedrock → Anthropic | Throughput (req/s) | 4,633 | concurrency 16; n=27,807 | Unbounded |
| OpenAI → Gemini | Added latency p50 (µs) | 626 | Valid · n=8,969 gateway / 262,364 direct | — |
| OpenAI → Gemini | Added latency p99 (µs) | 1,018 | Valid · n=8,969 gateway / 262,364 direct | — |
| OpenAI → Gemini | Gateway p99 at concurrency 1 (µs) | 1,046 | Valid · n=8,969 | — |
| OpenAI → Gemini | Direct-to-mock p99 at concurrency 1 (µs) | 28 | Valid · n=262,364 | — |
| OpenAI → Gemini | CPU per request (µs) | 913.8 | Valid | — |
| OpenAI → Gemini | Requests per CPU-second | 1,094.3 | Valid | — |
| OpenAI → Gemini | Cost window concurrency | 8 | Valid | — |
| OpenAI → Gemini | Cost window core utilisation | 0.988 | Valid | — |
| OpenAI → Gemini | Cost window successful requests | 26,034 | Valid | — |
| OpenAI → Gemini | Cost window throughput (req/s) | 4,338 | Valid | — |
| OpenAI → Gemini | Cost window threads | 14 | Valid | — |
| OpenAI → Gemini | Non-voluntary context switches per request | 0 | Valid | — |
| OpenAI → Gemini | Cost window major page faults | 0 | Valid | — |
| OpenAI → Gemini | Added time to first event p50 (µs) | — | Untestable | — |
| OpenAI → Gemini | Added time to first event p99 (µs) | — | Untestable | — |
| OpenAI → Gemini | Added inter-frame gap p50 (µs) | — | Untestable | — |
| OpenAI → Gemini | Added inter-frame gap p99 (µs) | — | Untestable | — |
| OpenAI → Gemini | Streams sustained (frames/s) | — | Untestable | — |
| OpenAI → Gemini | Streams sustained | — | Untestable | — |
| OpenAI → Gemini | Idle RSS (MiB) | 162 | Valid | — |
| OpenAI → Gemini | Steady-state RSS (MiB) | 1,093.6 | Valid | — |
| OpenAI → Gemini | Recovered RSS (MiB) | 1,193.6 | Valid | — |
| OpenAI → Gemini | Peak RSS (MiB) | 1,199.4 | Valid | — |
| OpenAI → Gemini | Peak RSS high-water mark (MiB) | 1,200.1 | Valid | — |
| OpenAI → Gemini | Time to plateau (s) | — | Not measured | — |
| OpenAI → Gemini | Memory growth (MiB/min) | 73.3 | Valid | — |
| OpenAI → Gemini | Memory load duration (s) | 300 | Valid | — |
| OpenAI → Gemini | Memory shape under load | 1 | Valid | — |
| OpenAI → Gemini | Idle memory shape | — | Not measured | — |
| OpenAI → Gemini | Idle memory static | 1 | Valid | — |
| OpenAI → Gemini | Idle memory growth (MiB/min) | 0.381 | Valid | — |
| OpenAI → Gemini | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| OpenAI → Gemini | Throughput (req/s) | 4,251 | concurrency 4; n=25,514 | 5 ms |
| OpenAI → Gemini | Throughput (req/s) | 4,415 | concurrency 8; n=26,497 | 10 ms |
| OpenAI → Gemini | Throughput (req/s) | 4,467 | concurrency 16; n=26,809 | 50 ms |
| OpenAI → Gemini | Throughput (req/s) | 4,467 | concurrency 16; n=26,809 | 100 ms |
| OpenAI → Gemini | Throughput (req/s) | 4,467 | concurrency 16; n=26,809 | Unbounded |
| OpenAI Responses → Gemini | Added latency p50 (µs) | 667 | Valid · n=8,355 gateway / 244,434 direct | — |
| OpenAI Responses → Gemini | Added latency p99 (µs) | 1,139 | Valid · n=8,355 gateway / 244,434 direct | — |
| OpenAI Responses → Gemini | Gateway p99 at concurrency 1 (µs) | 1,169 | Valid · n=8,355 | — |
| OpenAI Responses → Gemini | Direct-to-mock p99 at concurrency 1 (µs) | 30 | Valid · n=244,434 | — |
| OpenAI Responses → Gemini | CPU per request (µs) | 958.6 | Valid | — |
| OpenAI Responses → Gemini | Requests per CPU-second | 1,043.2 | Valid | — |
| OpenAI Responses → Gemini | Cost window concurrency | 8 | Valid | — |
| OpenAI Responses → Gemini | Cost window core utilisation | 0.991 | Valid | — |
| OpenAI Responses → Gemini | Cost window successful requests | 24,891 | Valid | — |
| OpenAI Responses → Gemini | Cost window throughput (req/s) | 4,147 | Valid | — |
| OpenAI Responses → Gemini | Cost window threads | 14 | Valid | — |
| OpenAI Responses → Gemini | Non-voluntary context switches per request | 0 | Valid | — |
| OpenAI Responses → Gemini | Cost window major page faults | 0 | Valid | — |
| OpenAI Responses → Gemini | Added time to first event p50 (µs) | — | Untestable | — |
| OpenAI Responses → Gemini | Added time to first event p99 (µs) | — | Untestable | — |
| OpenAI Responses → Gemini | Added inter-frame gap p50 (µs) | — | Untestable | — |
| OpenAI Responses → Gemini | Added inter-frame gap p99 (µs) | — | Untestable | — |
| OpenAI Responses → Gemini | Streams sustained (frames/s) | — | Untestable | — |
| OpenAI Responses → Gemini | Streams sustained | — | Untestable | — |
| OpenAI Responses → Gemini | Idle RSS (MiB) | 161 | Valid | — |
| OpenAI Responses → Gemini | Steady-state RSS (MiB) | 1,065.3 | Valid | — |
| OpenAI Responses → Gemini | Recovered RSS (MiB) | 1,112.7 | Valid | — |
| OpenAI Responses → Gemini | Peak RSS (MiB) | 1,287.9 | Valid | — |
| OpenAI Responses → Gemini | Peak RSS high-water mark (MiB) | 1,203.4 | Valid | — |
| OpenAI Responses → Gemini | Time to plateau (s) | — | Not measured | — |
| OpenAI Responses → Gemini | Memory growth (MiB/min) | 57.62 | Valid | — |
| OpenAI Responses → Gemini | Memory load duration (s) | 300 | Valid | — |
| OpenAI Responses → Gemini | Memory shape under load | 1 | Valid | — |
| OpenAI Responses → Gemini | Idle memory shape | — | Not measured | — |
| OpenAI Responses → Gemini | Idle memory static | 1 | Valid | — |
| OpenAI Responses → Gemini | Idle memory growth (MiB/min) | 0.163 | Valid | — |
| OpenAI Responses → Gemini | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| OpenAI Responses → Gemini | Throughput (req/s) | 4,033 | concurrency 4; n=24,211 | 5 ms |
| OpenAI Responses → Gemini | Throughput (req/s) | 4,174 | concurrency 8; n=25,048 | 10 ms |
| OpenAI Responses → Gemini | Throughput (req/s) | 4,222 | concurrency 16; n=25,335 | 50 ms |
| OpenAI Responses → Gemini | Throughput (req/s) | 4,222 | concurrency 16; n=25,335 | 100 ms |
| OpenAI Responses → Gemini | Throughput (req/s) | 4,222 | concurrency 16; n=25,335 | Unbounded |
| Anthropic → Gemini | Added latency p50 (µs) | 634 | Valid · n=8,607 gateway / 257,441 direct | — |
| Anthropic → Gemini | Added latency p99 (µs) | 1,125 | Valid · n=8,607 gateway / 257,441 direct | — |
| Anthropic → Gemini | Gateway p99 at concurrency 1 (µs) | 1,153 | Valid · n=8,607 | — |
| Anthropic → Gemini | Direct-to-mock p99 at concurrency 1 (µs) | 28 | Valid · n=257,441 | — |
| Anthropic → Gemini | CPU per request (µs) | 926.3 | Valid | — |
| Anthropic → Gemini | Requests per CPU-second | 1,079.6 | Valid | — |
| Anthropic → Gemini | Cost window concurrency | 8 | Valid | — |
| Anthropic → Gemini | Cost window core utilisation | 0.986 | Valid | — |
| Anthropic → Gemini | Cost window successful requests | 25,640 | Valid | — |
| Anthropic → Gemini | Cost window throughput (req/s) | 4,272 | Valid | — |
| Anthropic → Gemini | Cost window threads | 14 | Valid | — |
| Anthropic → Gemini | Non-voluntary context switches per request | 0.014 | Valid | — |
| Anthropic → Gemini | Cost window major page faults | 0 | Valid | — |
| Anthropic → Gemini | Added time to first event p50 (µs) | — | Untestable | — |
| Anthropic → Gemini | Added time to first event p99 (µs) | — | Untestable | — |
| Anthropic → Gemini | Added inter-frame gap p50 (µs) | — | Untestable | — |
| Anthropic → Gemini | Added inter-frame gap p99 (µs) | — | Untestable | — |
| Anthropic → Gemini | Streams sustained (frames/s) | — | Untestable | — |
| Anthropic → Gemini | Streams sustained | — | Untestable | — |
| Anthropic → Gemini | Idle RSS (MiB) | 160.8 | Valid | — |
| Anthropic → Gemini | Steady-state RSS (MiB) | 1,083.5 | Valid | — |
| Anthropic → Gemini | Recovered RSS (MiB) | 1,179.1 | Valid | — |
| Anthropic → Gemini | Peak RSS (MiB) | 1,239.1 | Valid | — |
| Anthropic → Gemini | Peak RSS high-water mark (MiB) | 1,228 | Valid | — |
| Anthropic → Gemini | Time to plateau (s) | — | Not measured | — |
| Anthropic → Gemini | Memory growth (MiB/min) | 81.53 | Valid | — |
| Anthropic → Gemini | Memory load duration (s) | 300 | Valid | — |
| Anthropic → Gemini | Memory shape under load | 1 | Valid | — |
| Anthropic → Gemini | Idle memory shape | — | Not measured | — |
| Anthropic → Gemini | Idle memory static | 1 | Valid | — |
| Anthropic → Gemini | Idle memory growth (MiB/min) | 0.711 | Valid | — |
| Anthropic → Gemini | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| Anthropic → Gemini | Throughput (req/s) | 4,055 | concurrency 4; n=24,334 | 5 ms |
| Anthropic → Gemini | Throughput (req/s) | 4,182 | concurrency 8; n=25,101 | 10 ms |
| Anthropic → Gemini | Throughput (req/s) | 4,257 | concurrency 16; n=25,547 | 50 ms |
| Anthropic → Gemini | Throughput (req/s) | 4,257 | concurrency 16; n=25,547 | 100 ms |
| Anthropic → Gemini | Throughput (req/s) | 4,257 | concurrency 16; n=25,547 | Unbounded |
| Gemini → Gemini | Added latency p50 (µs) | 652 | Valid · n=8,589 gateway / 266,124 direct | — |
| Gemini → Gemini | Added latency p99 (µs) | 1,098 | Valid · n=8,589 gateway / 266,124 direct | — |
| Gemini → Gemini | Gateway p99 at concurrency 1 (µs) | 1,125 | Valid · n=8,589 | — |
| Gemini → Gemini | Direct-to-mock p99 at concurrency 1 (µs) | 27 | Valid · n=266,124 | — |
| Gemini → Gemini | CPU per request (µs) | 954.9 | Valid | — |
| Gemini → Gemini | Requests per CPU-second | 1,047.2 | Valid | — |
| Gemini → Gemini | Cost window concurrency | 8 | Valid | — |
| Gemini → Gemini | Cost window core utilisation | 0.987 | Valid | — |
| Gemini → Gemini | Cost window successful requests | 24,871 | Valid | — |
| Gemini → Gemini | Cost window throughput (req/s) | 4,144 | Valid | — |
| Gemini → Gemini | Cost window threads | 14 | Valid | — |
| Gemini → Gemini | Non-voluntary context switches per request | 0 | Valid | — |
| Gemini → Gemini | Cost window major page faults | 0 | Valid | — |
| Gemini → Gemini | Added time to first event p50 (µs) | — | Untestable | — |
| Gemini → Gemini | Added time to first event p99 (µs) | — | Untestable | — |
| Gemini → Gemini | Added inter-frame gap p50 (µs) | — | Untestable | — |
| Gemini → Gemini | Added inter-frame gap p99 (µs) | — | Untestable | — |
| Gemini → Gemini | Streams sustained (frames/s) | — | Untestable | — |
| Gemini → Gemini | Streams sustained | — | Untestable | — |
| Gemini → Gemini | Idle RSS (MiB) | 161.5 | Valid | — |
| Gemini → Gemini | Steady-state RSS (MiB) | 1,066.1 | Valid | — |
| Gemini → Gemini | Recovered RSS (MiB) | 863.8 | Valid | — |
| Gemini → Gemini | Peak RSS (MiB) | 1,200.6 | Valid | — |
| Gemini → Gemini | Peak RSS high-water mark (MiB) | 1,200.7 | Valid | — |
| Gemini → Gemini | Time to plateau (s) | — | Not measured | — |
| Gemini → Gemini | Memory growth (MiB/min) | 86.41 | Valid | — |
| Gemini → Gemini | Memory load duration (s) | 300 | Valid | — |
| Gemini → Gemini | Memory shape under load | 1 | Valid | — |
| Gemini → Gemini | Idle memory shape | — | Not measured | — |
| Gemini → Gemini | Idle memory static | 1 | Valid | — |
| Gemini → Gemini | Idle memory growth (MiB/min) | 0.651 | Valid | — |
| Gemini → Gemini | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| Gemini → Gemini | Throughput (req/s) | 3,959 | concurrency 4; n=23,762 | 5 ms |
| Gemini → Gemini | Throughput (req/s) | 4,109 | concurrency 8; n=24,661 | 10 ms |
| Gemini → Gemini | Throughput (req/s) | 4,162 | concurrency 16; n=24,977 | 50 ms |
| Gemini → Gemini | Throughput (req/s) | 4,162 | concurrency 16; n=24,977 | 100 ms |
| Gemini → Gemini | Throughput (req/s) | 4,162 | concurrency 16; n=24,977 | Unbounded |
| Cohere → Gemini | Added latency p50 (µs) | 658 | Valid · n=8,497 gateway / 243,412 direct | — |
| Cohere → Gemini | Added latency p99 (µs) | 1,090 | Valid · n=8,497 gateway / 243,412 direct | — |
| Cohere → Gemini | Gateway p99 at concurrency 1 (µs) | 1,120 | Valid · n=8,497 | — |
| Cohere → Gemini | Direct-to-mock p99 at concurrency 1 (µs) | 30 | Valid · n=243,412 | — |
| Cohere → Gemini | CPU per request (µs) | 953.7 | Valid | — |
| Cohere → Gemini | Requests per CPU-second | 1,048.6 | Valid | — |
| Cohere → Gemini | Cost window concurrency | 8 | Valid | — |
| Cohere → Gemini | Cost window core utilisation | 0.989 | Valid | — |
| Cohere → Gemini | Cost window successful requests | 24,966 | Valid | — |
| Cohere → Gemini | Cost window throughput (req/s) | 4,160 | Valid | — |
| Cohere → Gemini | Cost window threads | 15 | Valid | — |
| Cohere → Gemini | Non-voluntary context switches per request | 0 | Valid | — |
| Cohere → Gemini | Cost window major page faults | 0 | Valid | — |
| Cohere → Gemini | Added time to first event p50 (µs) | — | Untestable | — |
| Cohere → Gemini | Added time to first event p99 (µs) | — | Untestable | — |
| Cohere → Gemini | Added inter-frame gap p50 (µs) | — | Untestable | — |
| Cohere → Gemini | Added inter-frame gap p99 (µs) | — | Untestable | — |
| Cohere → Gemini | Streams sustained (frames/s) | — | Untestable | — |
| Cohere → Gemini | Streams sustained | — | Untestable | — |
| Cohere → Gemini | Idle RSS (MiB) | 160.1 | Valid | — |
| Cohere → Gemini | Steady-state RSS (MiB) | 986.2 | Valid | — |
| Cohere → Gemini | Recovered RSS (MiB) | 1,046.5 | Valid | — |
| Cohere → Gemini | Peak RSS (MiB) | 1,179.5 | Valid | — |
| Cohere → Gemini | Peak RSS high-water mark (MiB) | 1,142.3 | Valid | — |
| Cohere → Gemini | Time to plateau (s) | — | Not measured | — |
| Cohere → Gemini | Memory growth (MiB/min) | 69.08 | Valid | — |
| Cohere → Gemini | Memory load duration (s) | 300 | Valid | — |
| Cohere → Gemini | Memory shape under load | 1 | Valid | — |
| Cohere → Gemini | Idle memory shape | — | Not measured | — |
| Cohere → Gemini | Idle memory static | 1 | Valid | — |
| Cohere → Gemini | Idle memory growth (MiB/min) | 0.439 | Valid | — |
| Cohere → Gemini | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| Cohere → Gemini | Throughput (req/s) | 4,011 | concurrency 4; n=24,072 | 5 ms |
| Cohere → Gemini | Throughput (req/s) | 4,196 | concurrency 8; n=25,181 | 10 ms |
| Cohere → Gemini | Throughput (req/s) | 4,233 | concurrency 16; n=25,400 | 50 ms |
| Cohere → Gemini | Throughput (req/s) | 4,233 | concurrency 16; n=25,400 | 100 ms |
| Cohere → Gemini | Throughput (req/s) | 4,233 | concurrency 16; n=25,400 | Unbounded |
| Bedrock → Gemini | Added latency p50 (µs) | 607 | Valid · n=9,211 gateway / 248,434 direct | — |
| Bedrock → Gemini | Added latency p99 (µs) | 1,009 | Valid · n=9,211 gateway / 248,434 direct | — |
| Bedrock → Gemini | Gateway p99 at concurrency 1 (µs) | 1,038 | Valid · n=9,211 | — |
| Bedrock → Gemini | Direct-to-mock p99 at concurrency 1 (µs) | 29 | Valid · n=248,434 | — |
| Bedrock → Gemini | CPU per request (µs) | 898 | Valid | — |
| Bedrock → Gemini | Requests per CPU-second | 1,113.6 | Valid | — |
| Bedrock → Gemini | Cost window concurrency | 8 | Valid | — |
| Bedrock → Gemini | Cost window core utilisation | 0.993 | Valid | — |
| Bedrock → Gemini | Cost window successful requests | 26,614 | Valid | — |
| Bedrock → Gemini | Cost window throughput (req/s) | 4,435 | Valid | — |
| Bedrock → Gemini | Cost window threads | 13 | Valid | — |
| Bedrock → Gemini | Non-voluntary context switches per request | 0 | Valid | — |
| Bedrock → Gemini | Cost window major page faults | 0 | Valid | — |
| Bedrock → Gemini | Added time to first event p50 (µs) | — | Untestable | — |
| Bedrock → Gemini | Added time to first event p99 (µs) | — | Untestable | — |
| Bedrock → Gemini | Added inter-frame gap p50 (µs) | — | Untestable | — |
| Bedrock → Gemini | Added inter-frame gap p99 (µs) | — | Untestable | — |
| Bedrock → Gemini | Streams sustained (frames/s) | — | Untestable | — |
| Bedrock → Gemini | Streams sustained | — | Untestable | — |
| Bedrock → Gemini | Idle RSS (MiB) | 160.7 | Valid | — |
| Bedrock → Gemini | Steady-state RSS (MiB) | 1,057.9 | Valid | — |
| Bedrock → Gemini | Recovered RSS (MiB) | 928.5 | Valid | — |
| Bedrock → Gemini | Peak RSS (MiB) | 1,270.6 | Valid | — |
| Bedrock → Gemini | Peak RSS high-water mark (MiB) | 1,157.7 | Valid | — |
| Bedrock → Gemini | Time to plateau (s) | — | Not measured | — |
| Bedrock → Gemini | Memory growth (MiB/min) | 64.94 | Valid | — |
| Bedrock → Gemini | Memory load duration (s) | 300 | Valid | — |
| Bedrock → Gemini | Memory shape under load | 1 | Valid | — |
| Bedrock → Gemini | Idle memory shape | — | Not measured | — |
| Bedrock → Gemini | Idle memory static | 1 | Valid | — |
| Bedrock → Gemini | Idle memory growth (MiB/min) | 0.335 | Valid | — |
| Bedrock → Gemini | Throughput (req/s) | 1,551 | concurrency 1; n=9,310 | 1 ms |
| Bedrock → Gemini | Throughput (req/s) | 4,311 | concurrency 4; n=25,871 | 5 ms |
| Bedrock → Gemini | Throughput (req/s) | 4,413 | concurrency 8; n=26,482 | 10 ms |
| Bedrock → Gemini | Throughput (req/s) | 4,603 | concurrency 16; n=27,624 | 50 ms |
| Bedrock → Gemini | Throughput (req/s) | 4,603 | concurrency 16; n=27,624 | 100 ms |
| Bedrock → Gemini | Throughput (req/s) | 4,603 | concurrency 16; n=27,624 | Unbounded |
| OpenAI → Cohere | Added latency p50 (µs) | 641 | Valid · n=8,677 gateway / 263,654 direct | — |
| OpenAI → Cohere | Added latency p99 (µs) | 1,090 | Valid · n=8,677 gateway / 263,654 direct | — |
| OpenAI → Cohere | Gateway p99 at concurrency 1 (µs) | 1,118 | Valid · n=8,677 | — |
| OpenAI → Cohere | Direct-to-mock p99 at concurrency 1 (µs) | 28 | Valid · n=263,654 | — |
| OpenAI → Cohere | CPU per request (µs) | 900.1 | Valid | — |
| OpenAI → Cohere | Requests per CPU-second | 1,111 | Valid | — |
| OpenAI → Cohere | Cost window concurrency | 8 | Valid | — |
| OpenAI → Cohere | Cost window core utilisation | 0.991 | Valid | — |
| OpenAI → Cohere | Cost window successful requests | 26,498 | Valid | — |
| OpenAI → Cohere | Cost window throughput (req/s) | 4,415 | Valid | — |
| OpenAI → Cohere | Cost window threads | 14 | Valid | — |
| OpenAI → Cohere | Non-voluntary context switches per request | 0 | Valid | — |
| OpenAI → Cohere | Cost window major page faults | 0 | Valid | — |
| OpenAI → Cohere | Added time to first event p50 (µs) | — | Untestable | — |
| OpenAI → Cohere | Added time to first event p99 (µs) | — | Untestable | — |
| OpenAI → Cohere | Added inter-frame gap p50 (µs) | — | Untestable | — |
| OpenAI → Cohere | Added inter-frame gap p99 (µs) | — | Untestable | — |
| OpenAI → Cohere | Streams sustained (frames/s) | — | Untestable | — |
| OpenAI → Cohere | Streams sustained | — | Untestable | — |
| OpenAI → Cohere | Idle RSS (MiB) | 161.4 | Valid | — |
| OpenAI → Cohere | Steady-state RSS (MiB) | 1,119.1 | Valid | — |
| OpenAI → Cohere | Recovered RSS (MiB) | 1,071.8 | Valid | — |
| OpenAI → Cohere | Peak RSS (MiB) | 1,254.9 | Valid | — |
| OpenAI → Cohere | Peak RSS high-water mark (MiB) | 1,211.2 | Valid | — |
| OpenAI → Cohere | Time to plateau (s) | — | Not measured | — |
| OpenAI → Cohere | Memory growth (MiB/min) | 74.02 | Valid | — |
| OpenAI → Cohere | Memory load duration (s) | 300 | Valid | — |
| OpenAI → Cohere | Memory shape under load | 1 | Valid | — |
| OpenAI → Cohere | Idle memory shape | — | Not measured | — |
| OpenAI → Cohere | Idle memory static | 1 | Valid | — |
| OpenAI → Cohere | Idle memory growth (MiB/min) | 0.156 | Valid | — |
| OpenAI → Cohere | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| OpenAI → Cohere | Throughput (req/s) | 4,074 | concurrency 4; n=24,453 | 5 ms |
| OpenAI → Cohere | Throughput (req/s) | 4,339 | concurrency 8; n=26,039 | 10 ms |
| OpenAI → Cohere | Throughput (req/s) | 4,507 | concurrency 32; n=27,048 | 50 ms |
| OpenAI → Cohere | Throughput (req/s) | 4,507 | concurrency 32; n=27,048 | 100 ms |
| OpenAI → Cohere | Throughput (req/s) | 4,507 | concurrency 32; n=27,048 | Unbounded |
| OpenAI Responses → Cohere | Added latency p50 (µs) | 630 | Valid · n=8,900 gateway / 251,031 direct | — |
| OpenAI Responses → Cohere | Added latency p99 (µs) | 1,032 | Valid · n=8,900 gateway / 251,031 direct | — |
| OpenAI Responses → Cohere | Gateway p99 at concurrency 1 (µs) | 1,061 | Valid · n=8,900 | — |
| OpenAI Responses → Cohere | Direct-to-mock p99 at concurrency 1 (µs) | 29 | Valid · n=251,031 | — |
| OpenAI Responses → Cohere | CPU per request (µs) | 907.9 | Valid | — |
| OpenAI Responses → Cohere | Requests per CPU-second | 1,101.4 | Valid | — |
| OpenAI Responses → Cohere | Cost window concurrency | 8 | Valid | — |
| OpenAI Responses → Cohere | Cost window core utilisation | 0.987 | Valid | — |
| OpenAI Responses → Cohere | Cost window successful requests | 26,169 | Valid | — |
| OpenAI Responses → Cohere | Cost window throughput (req/s) | 4,360 | Valid | — |
| OpenAI Responses → Cohere | Cost window threads | 14 | Valid | — |
| OpenAI Responses → Cohere | Non-voluntary context switches per request | 0.037 | Valid | — |
| OpenAI Responses → Cohere | Cost window major page faults | 0 | Valid | — |
| OpenAI Responses → Cohere | Added time to first event p50 (µs) | — | Untestable | — |
| OpenAI Responses → Cohere | Added time to first event p99 (µs) | — | Untestable | — |
| OpenAI Responses → Cohere | Added inter-frame gap p50 (µs) | — | Untestable | — |
| OpenAI Responses → Cohere | Added inter-frame gap p99 (µs) | — | Untestable | — |
| OpenAI Responses → Cohere | Streams sustained (frames/s) | — | Untestable | — |
| OpenAI Responses → Cohere | Streams sustained | — | Untestable | — |
| OpenAI Responses → Cohere | Idle RSS (MiB) | 161.4 | Valid | — |
| OpenAI Responses → Cohere | Steady-state RSS (MiB) | 1,074.9 | Valid | — |
| OpenAI Responses → Cohere | Recovered RSS (MiB) | 1,037.1 | Valid | — |
| OpenAI Responses → Cohere | Peak RSS (MiB) | 1,256.9 | Valid | — |
| OpenAI Responses → Cohere | Peak RSS high-water mark (MiB) | 1,184.6 | Valid | — |
| OpenAI Responses → Cohere | Time to plateau (s) | — | Not measured | — |
| OpenAI Responses → Cohere | Memory growth (MiB/min) | 79.79 | Valid | — |
| OpenAI Responses → Cohere | Memory load duration (s) | 300 | Valid | — |
| OpenAI Responses → Cohere | Memory shape under load | 1 | Valid | — |
| OpenAI Responses → Cohere | Idle memory shape | — | Not measured | — |
| OpenAI Responses → Cohere | Idle memory static | 1 | Valid | — |
| OpenAI Responses → Cohere | Idle memory growth (MiB/min) | 0.51 | Valid | — |
| OpenAI Responses → Cohere | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| OpenAI Responses → Cohere | Throughput (req/s) | 4,135 | concurrency 4; n=24,816 | 5 ms |
| OpenAI Responses → Cohere | Throughput (req/s) | 4,290 | concurrency 8; n=25,751 | 10 ms |
| OpenAI Responses → Cohere | Throughput (req/s) | 4,330 | concurrency 16; n=25,988 | 50 ms |
| OpenAI Responses → Cohere | Throughput (req/s) | 4,330 | concurrency 16; n=25,988 | 100 ms |
| OpenAI Responses → Cohere | Throughput (req/s) | 4,330 | concurrency 16; n=25,988 | Unbounded |
| Anthropic → Cohere | Added latency p50 (µs) | 608 | Valid · n=9,106 gateway / 260,492 direct | — |
| Anthropic → Cohere | Added latency p99 (µs) | 1,061 | Valid · n=9,106 gateway / 260,492 direct | — |
| Anthropic → Cohere | Gateway p99 at concurrency 1 (µs) | 1,089 | Valid · n=9,106 | — |
| Anthropic → Cohere | Direct-to-mock p99 at concurrency 1 (µs) | 28 | Valid · n=260,492 | — |
| Anthropic → Cohere | CPU per request (µs) | 892.7 | Valid | — |
| Anthropic → Cohere | Requests per CPU-second | 1,120.2 | Valid | — |
| Anthropic → Cohere | Cost window concurrency | 8 | Valid | — |
| Anthropic → Cohere | Cost window core utilisation | 0.991 | Valid | — |
| Anthropic → Cohere | Cost window successful requests | 26,751 | Valid | — |
| Anthropic → Cohere | Cost window throughput (req/s) | 4,457 | Valid | — |
| Anthropic → Cohere | Cost window threads | 14 | Valid | — |
| Anthropic → Cohere | Non-voluntary context switches per request | 0.018 | Valid | — |
| Anthropic → Cohere | Cost window major page faults | 0 | Valid | — |
| Anthropic → Cohere | Added time to first event p50 (µs) | — | Untestable | — |
| Anthropic → Cohere | Added time to first event p99 (µs) | — | Untestable | — |
| Anthropic → Cohere | Added inter-frame gap p50 (µs) | — | Untestable | — |
| Anthropic → Cohere | Added inter-frame gap p99 (µs) | — | Untestable | — |
| Anthropic → Cohere | Streams sustained (frames/s) | — | Untestable | — |
| Anthropic → Cohere | Streams sustained | — | Untestable | — |
| Anthropic → Cohere | Idle RSS (MiB) | 161.3 | Valid | — |
| Anthropic → Cohere | Steady-state RSS (MiB) | 1,118.1 | Valid | — |
| Anthropic → Cohere | Recovered RSS (MiB) | 1,140.1 | Valid | — |
| Anthropic → Cohere | Peak RSS (MiB) | 1,222.5 | Valid | — |
| Anthropic → Cohere | Peak RSS high-water mark (MiB) | 1,216.4 | Valid | — |
| Anthropic → Cohere | Time to plateau (s) | — | Not measured | — |
| Anthropic → Cohere | Memory growth (MiB/min) | 68.59 | Valid | — |
| Anthropic → Cohere | Memory load duration (s) | 300 | Valid | — |
| Anthropic → Cohere | Memory shape under load | 1 | Valid | — |
| Anthropic → Cohere | Idle memory shape | — | Not measured | — |
| Anthropic → Cohere | Idle memory static | 1 | Valid | — |
| Anthropic → Cohere | Idle memory growth (MiB/min) | 0.167 | Valid | — |
| Anthropic → Cohere | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| Anthropic → Cohere | Throughput (req/s) | 4,182 | concurrency 4; n=25,097 | 5 ms |
| Anthropic → Cohere | Throughput (req/s) | 4,364 | concurrency 8; n=26,188 | 10 ms |
| Anthropic → Cohere | Throughput (req/s) | 4,404 | concurrency 16; n=26,429 | 50 ms |
| Anthropic → Cohere | Throughput (req/s) | 4,404 | concurrency 16; n=26,429 | 100 ms |
| Anthropic → Cohere | Throughput (req/s) | 4,404 | concurrency 16; n=26,429 | Unbounded |
| Gemini → Cohere | Added latency p50 (µs) | 670 | Valid · n=8,377 gateway / 268,876 direct | — |
| Gemini → Cohere | Added latency p99 (µs) | 1,111 | Valid · n=8,377 gateway / 268,876 direct | — |
| Gemini → Cohere | Gateway p99 at concurrency 1 (µs) | 1,138 | Valid · n=8,377 | — |
| Gemini → Cohere | Direct-to-mock p99 at concurrency 1 (µs) | 27 | Valid · n=268,876 | — |
| Gemini → Cohere | CPU per request (µs) | 944.4 | Valid | — |
| Gemini → Cohere | Requests per CPU-second | 1,058.8 | Valid | — |
| Gemini → Cohere | Cost window concurrency | 8 | Valid | — |
| Gemini → Cohere | Cost window core utilisation | 0.989 | Valid | — |
| Gemini → Cohere | Cost window successful requests | 25,211 | Valid | — |
| Gemini → Cohere | Cost window throughput (req/s) | 4,200 | Valid | — |
| Gemini → Cohere | Cost window threads | 14 | Valid | — |
| Gemini → Cohere | Non-voluntary context switches per request | 0.046 | Valid | — |
| Gemini → Cohere | Cost window major page faults | 0 | Valid | — |
| Gemini → Cohere | Added time to first event p50 (µs) | — | Untestable | — |
| Gemini → Cohere | Added time to first event p99 (µs) | — | Untestable | — |
| Gemini → Cohere | Added inter-frame gap p50 (µs) | — | Untestable | — |
| Gemini → Cohere | Added inter-frame gap p99 (µs) | — | Untestable | — |
| Gemini → Cohere | Streams sustained (frames/s) | — | Untestable | — |
| Gemini → Cohere | Streams sustained | — | Untestable | — |
| Gemini → Cohere | Idle RSS (MiB) | 162.3 | Valid | — |
| Gemini → Cohere | Steady-state RSS (MiB) | 1,063 | Valid | — |
| Gemini → Cohere | Recovered RSS (MiB) | 1,089.3 | Valid | — |
| Gemini → Cohere | Peak RSS (MiB) | 1,247.8 | Valid | — |
| Gemini → Cohere | Peak RSS high-water mark (MiB) | 1,208.6 | Valid | — |
| Gemini → Cohere | Time to plateau (s) | — | Not measured | — |
| Gemini → Cohere | Memory growth (MiB/min) | 89.15 | Valid | — |
| Gemini → Cohere | Memory load duration (s) | 300 | Valid | — |
| Gemini → Cohere | Memory shape under load | 1 | Valid | — |
| Gemini → Cohere | Idle memory shape | — | Not measured | — |
| Gemini → Cohere | Idle memory static | 1 | Valid | — |
| Gemini → Cohere | Idle memory growth (MiB/min) | 0.289 | Valid | — |
| Gemini → Cohere | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| Gemini → Cohere | Throughput (req/s) | 4,019 | concurrency 4; n=24,118 | 5 ms |
| Gemini → Cohere | Throughput (req/s) | 4,163 | concurrency 8; n=24,987 | 10 ms |
| Gemini → Cohere | Throughput (req/s) | 4,191 | concurrency 16; n=25,155 | 50 ms |
| Gemini → Cohere | Throughput (req/s) | 4,191 | concurrency 16; n=25,155 | 100 ms |
| Gemini → Cohere | Throughput (req/s) | 4,191 | concurrency 16; n=25,155 | Unbounded |
| Cohere → Cohere | Added latency p50 (µs) | 609 | Valid · n=9,094 gateway / 264,951 direct | — |
| Cohere → Cohere | Added latency p99 (µs) | 1,062 | Valid · n=9,094 gateway / 264,951 direct | — |
| Cohere → Cohere | Gateway p99 at concurrency 1 (µs) | 1,090 | Valid · n=9,094 | — |
| Cohere → Cohere | Direct-to-mock p99 at concurrency 1 (µs) | 28 | Valid · n=264,951 | — |
| Cohere → Cohere | CPU per request (µs) | 888.4 | Valid | — |
| Cohere → Cohere | Requests per CPU-second | 1,125.7 | Valid | — |
| Cohere → Cohere | Cost window concurrency | 8 | Valid | — |
| Cohere → Cohere | Cost window core utilisation | 0.992 | Valid | — |
| Cohere → Cohere | Cost window successful requests | 26,881 | Valid | — |
| Cohere → Cohere | Cost window throughput (req/s) | 4,479 | Valid | — |
| Cohere → Cohere | Cost window threads | 14 | Valid | — |
| Cohere → Cohere | Non-voluntary context switches per request | 0.019 | Valid | — |
| Cohere → Cohere | Cost window major page faults | 0 | Valid | — |
| Cohere → Cohere | Added time to first event p50 (µs) | — | Untestable | — |
| Cohere → Cohere | Added time to first event p99 (µs) | — | Untestable | — |
| Cohere → Cohere | Added inter-frame gap p50 (µs) | — | Untestable | — |
| Cohere → Cohere | Added inter-frame gap p99 (µs) | — | Untestable | — |
| Cohere → Cohere | Streams sustained (frames/s) | — | Untestable | — |
| Cohere → Cohere | Streams sustained | — | Untestable | — |
| Cohere → Cohere | Idle RSS (MiB) | 160.7 | Valid | — |
| Cohere → Cohere | Steady-state RSS (MiB) | 1,074.5 | Valid | — |
| Cohere → Cohere | Recovered RSS (MiB) | 1,027.6 | Valid | — |
| Cohere → Cohere | Peak RSS (MiB) | 1,245.7 | Valid | — |
| Cohere → Cohere | Peak RSS high-water mark (MiB) | 1,159 | Valid | — |
| Cohere → Cohere | Time to plateau (s) | — | Not measured | — |
| Cohere → Cohere | Memory growth (MiB/min) | 76.4 | Valid | — |
| Cohere → Cohere | Memory load duration (s) | 300 | Valid | — |
| Cohere → Cohere | Memory shape under load | 1 | Valid | — |
| Cohere → Cohere | Idle memory shape | — | Not measured | — |
| Cohere → Cohere | Idle memory static | 1 | Valid | — |
| Cohere → Cohere | Idle memory growth (MiB/min) | 0.628 | Valid | — |
| Cohere → Cohere | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| Cohere → Cohere | Throughput (req/s) | 4,379 | concurrency 4; n=26,284 | 5 ms |
| Cohere → Cohere | Throughput (req/s) | 4,498 | concurrency 8; n=26,992 | 10 ms |
| Cohere → Cohere | Throughput (req/s) | 4,544 | concurrency 16; n=27,267 | 50 ms |
| Cohere → Cohere | Throughput (req/s) | 4,544 | concurrency 16; n=27,267 | 100 ms |
| Cohere → Cohere | Throughput (req/s) | 4,544 | concurrency 16; n=27,267 | Unbounded |
| Bedrock → Cohere | Added latency p50 (µs) | 553 | Valid · n=10,081 gateway / 263,966 direct | — |
| Bedrock → Cohere | Added latency p99 (µs) | 938 | Valid · n=10,081 gateway / 263,966 direct | — |
| Bedrock → Cohere | Gateway p99 at concurrency 1 (µs) | 966 | Valid · n=10,081 | — |
| Bedrock → Cohere | Direct-to-mock p99 at concurrency 1 (µs) | 28 | Valid · n=263,966 | — |
| Bedrock → Cohere | CPU per request (µs) | 808.9 | Valid | — |
| Bedrock → Cohere | Requests per CPU-second | 1,236.3 | Valid | — |
| Bedrock → Cohere | Cost window concurrency | 8 | Valid | — |
| Bedrock → Cohere | Cost window core utilisation | 0.99 | Valid | — |
| Bedrock → Cohere | Cost window successful requests | 29,461 | Valid | — |
| Bedrock → Cohere | Cost window throughput (req/s) | 4,909 | Valid | — |
| Bedrock → Cohere | Cost window threads | 14 | Valid | — |
| Bedrock → Cohere | Non-voluntary context switches per request | 0.016 | Valid | — |
| Bedrock → Cohere | Cost window major page faults | 0 | Valid | — |
| Bedrock → Cohere | Added time to first event p50 (µs) | — | Untestable | — |
| Bedrock → Cohere | Added time to first event p99 (µs) | — | Untestable | — |
| Bedrock → Cohere | Added inter-frame gap p50 (µs) | — | Untestable | — |
| Bedrock → Cohere | Added inter-frame gap p99 (µs) | — | Untestable | — |
| Bedrock → Cohere | Streams sustained (frames/s) | — | Untestable | — |
| Bedrock → Cohere | Streams sustained | — | Untestable | — |
| Bedrock → Cohere | Idle RSS (MiB) | 161.3 | Valid | — |
| Bedrock → Cohere | Steady-state RSS (MiB) | 1,081.5 | Valid | — |
| Bedrock → Cohere | Recovered RSS (MiB) | 1,024 | Valid | — |
| Bedrock → Cohere | Peak RSS (MiB) | 1,293.4 | Valid | — |
| Bedrock → Cohere | Peak RSS high-water mark (MiB) | 1,176.1 | Valid | — |
| Bedrock → Cohere | Time to plateau (s) | — | Not measured | — |
| Bedrock → Cohere | Memory growth (MiB/min) | 70.1 | Valid | — |
| Bedrock → Cohere | Memory load duration (s) | 300 | Valid | — |
| Bedrock → Cohere | Memory shape under load | 1 | Valid | — |
| Bedrock → Cohere | Idle memory shape | — | Not measured | — |
| Bedrock → Cohere | Idle memory static | 1 | Valid | — |
| Bedrock → Cohere | Idle memory growth (MiB/min) | 0.335 | Valid | — |
| Bedrock → Cohere | Throughput (req/s) | 1,683 | concurrency 1; n=10,101 | 1 ms |
| Bedrock → Cohere | Throughput (req/s) | 4,732 | concurrency 4; n=28,401 | 5 ms |
| Bedrock → Cohere | Throughput (req/s) | 4,965 | concurrency 8; n=29,795 | 10 ms |
| Bedrock → Cohere | Throughput (req/s) | 5,048 | concurrency 16; n=30,292 | 50 ms |
| Bedrock → Cohere | Throughput (req/s) | 5,048 | concurrency 16; n=30,292 | 100 ms |
| Bedrock → Cohere | Throughput (req/s) | 5,048 | concurrency 16; n=30,292 | Unbounded |
| OpenAI → Bedrock (upstream over HTTPS) | Added latency p50 (µs) | 708 | Valid · n=7,893 gateway / 260,423 direct | — |
| OpenAI → Bedrock (upstream over HTTPS) | Added latency p99 (µs) | 1,098 | Valid · n=7,893 gateway / 260,423 direct | — |
| OpenAI → Bedrock (upstream over HTTPS) | Gateway p99 at concurrency 1 (µs) | 1,126 | Valid · n=7,893 | — |
| OpenAI → Bedrock (upstream over HTTPS) | Direct-to-mock p99 at concurrency 1 (µs) | 28 | Valid · n=260,423 | — |
| OpenAI → Bedrock (upstream over HTTPS) | CPU per request (µs) | 955.8 | Valid | — |
| OpenAI → Bedrock (upstream over HTTPS) | Requests per CPU-second | 1,046.2 | Valid | — |
| OpenAI → Bedrock (upstream over HTTPS) | Cost window concurrency | 8 | Valid | — |
| OpenAI → Bedrock (upstream over HTTPS) | Cost window core utilisation | 0.989 | Valid | — |
| OpenAI → Bedrock (upstream over HTTPS) | Cost window successful requests | 24,910 | Valid | — |
| OpenAI → Bedrock (upstream over HTTPS) | Cost window throughput (req/s) | 4,151 | Valid | — |
| OpenAI → Bedrock (upstream over HTTPS) | Cost window threads | 17 | Valid | — |
| OpenAI → Bedrock (upstream over HTTPS) | Non-voluntary context switches per request | 0 | Valid | — |
| OpenAI → Bedrock (upstream over HTTPS) | Cost window major page faults | 0 | Valid | — |
| OpenAI → Bedrock (upstream over HTTPS) | Added time to first event p50 (µs) | — | Untestable | — |
| OpenAI → Bedrock (upstream over HTTPS) | Added time to first event p99 (µs) | — | Untestable | — |
| OpenAI → Bedrock (upstream over HTTPS) | Added inter-frame gap p50 (µs) | — | Untestable | — |
| OpenAI → Bedrock (upstream over HTTPS) | Added inter-frame gap p99 (µs) | — | Untestable | — |
| OpenAI → Bedrock (upstream over HTTPS) | Streams sustained (frames/s) | — | Untestable | — |
| OpenAI → Bedrock (upstream over HTTPS) | Streams sustained | — | Untestable | — |
| OpenAI → Bedrock (upstream over HTTPS) | Idle RSS (MiB) | 160 | Valid | — |
| OpenAI → Bedrock (upstream over HTTPS) | Steady-state RSS (MiB) | 981.9 | Valid | — |
| OpenAI → Bedrock (upstream over HTTPS) | Recovered RSS (MiB) | 1,138 | Valid | — |
| OpenAI → Bedrock (upstream over HTTPS) | Peak RSS (MiB) | 1,138.1 | Valid | — |
| OpenAI → Bedrock (upstream over HTTPS) | Peak RSS high-water mark (MiB) | 1,131 | Valid | — |
| OpenAI → Bedrock (upstream over HTTPS) | Time to plateau (s) | — | Not measured | — |
| OpenAI → Bedrock (upstream over HTTPS) | Memory growth (MiB/min) | 92.38 | Valid | — |
| OpenAI → Bedrock (upstream over HTTPS) | Memory load duration (s) | 300 | Valid | — |
| OpenAI → Bedrock (upstream over HTTPS) | Memory shape under load | 1 | Valid | — |
| OpenAI → Bedrock (upstream over HTTPS) | Idle memory shape | — | Not measured | — |
| OpenAI → Bedrock (upstream over HTTPS) | Idle memory static | 1 | Valid | — |
| OpenAI → Bedrock (upstream over HTTPS) | Idle memory growth (MiB/min) | 0.746 | Valid | — |
| OpenAI → Bedrock (upstream over HTTPS) | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| OpenAI → Bedrock (upstream over HTTPS) | Throughput (req/s) | 3,880 | concurrency 4; n=23,291 | 5 ms |
| OpenAI → Bedrock (upstream over HTTPS) | Throughput (req/s) | 4,031 | concurrency 8; n=24,194 | 10 ms |
| OpenAI → Bedrock (upstream over HTTPS) | Throughput (req/s) | 4,078 | concurrency 32; n=24,475 | 50 ms |
| OpenAI → Bedrock (upstream over HTTPS) | Throughput (req/s) | 4,078 | concurrency 32; n=24,475 | 100 ms |
| OpenAI → Bedrock (upstream over HTTPS) | Throughput (req/s) | 4,078 | concurrency 32; n=24,475 | Unbounded |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Added latency p50 (µs) | 728 | Valid · n=7,776 gateway / 260,884 direct | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Added latency p99 (µs) | 1,097 | Valid · n=7,776 gateway / 260,884 direct | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Gateway p99 at concurrency 1 (µs) | 1,125 | Valid · n=7,776 | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Direct-to-mock p99 at concurrency 1 (µs) | 28 | Valid · n=260,884 | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | CPU per request (µs) | 995.5 | Valid | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Requests per CPU-second | 1,004.6 | Valid | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Cost window concurrency | 8 | Valid | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Cost window core utilisation | 0.984 | Valid | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Cost window successful requests | 23,808 | Valid | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Cost window throughput (req/s) | 3,967 | Valid | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Cost window threads | 14 | Valid | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Non-voluntary context switches per request | 0 | Valid | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Cost window major page faults | 0 | Valid | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Added time to first event p50 (µs) | — | Untestable | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Added time to first event p99 (µs) | — | Untestable | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Added inter-frame gap p50 (µs) | — | Untestable | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Added inter-frame gap p99 (µs) | — | Untestable | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Streams sustained (frames/s) | — | Untestable | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Streams sustained | — | Untestable | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Idle RSS (MiB) | 161.1 | Valid | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Steady-state RSS (MiB) | 922.3 | Valid | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Recovered RSS (MiB) | 1,033 | Valid | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Peak RSS (MiB) | 1,145.9 | Valid | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Peak RSS high-water mark (MiB) | 1,102.1 | Valid | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Time to plateau (s) | — | Not measured | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Memory growth (MiB/min) | 104.2 | Valid | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Memory load duration (s) | 300 | Valid | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Memory shape under load | 1 | Valid | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Idle memory shape | — | Not measured | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Idle memory static | 1 | Valid | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Idle memory growth (MiB/min) | 0.265 | Valid | — |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Throughput (req/s) | 3,740 | concurrency 4; n=22,445 | 5 ms |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Throughput (req/s) | 3,865 | concurrency 8; n=23,201 | 10 ms |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Throughput (req/s) | 3,918 | concurrency 32; n=23,511 | 50 ms |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Throughput (req/s) | 3,918 | concurrency 32; n=23,511 | 100 ms |
| OpenAI Responses → Bedrock (upstream over HTTPS) | Throughput (req/s) | 3,918 | concurrency 32; n=23,511 | Unbounded |
| Anthropic → Bedrock (upstream over HTTPS) | Added latency p50 (µs) | 726 | Valid · n=7,777 gateway / 244,991 direct | — |
| Anthropic → Bedrock (upstream over HTTPS) | Added latency p99 (µs) | 1,099 | Valid · n=7,777 gateway / 244,991 direct | — |
| Anthropic → Bedrock (upstream over HTTPS) | Gateway p99 at concurrency 1 (µs) | 1,129 | Valid · n=7,777 | — |
| Anthropic → Bedrock (upstream over HTTPS) | Direct-to-mock p99 at concurrency 1 (µs) | 30 | Valid · n=244,991 | — |
| Anthropic → Bedrock (upstream over HTTPS) | CPU per request (µs) | 1,001.5 | Valid | — |
| Anthropic → Bedrock (upstream over HTTPS) | Requests per CPU-second | 998.5 | Valid | — |
| Anthropic → Bedrock (upstream over HTTPS) | Cost window concurrency | 8 | Valid | — |
| Anthropic → Bedrock (upstream over HTTPS) | Cost window core utilisation | 0.99 | Valid | — |
| Anthropic → Bedrock (upstream over HTTPS) | Cost window successful requests | 23,804 | Valid | — |
| Anthropic → Bedrock (upstream over HTTPS) | Cost window throughput (req/s) | 3,966 | Valid | — |
| Anthropic → Bedrock (upstream over HTTPS) | Cost window threads | 20 | Valid | — |
| Anthropic → Bedrock (upstream over HTTPS) | Non-voluntary context switches per request | 0.033 | Valid | — |
| Anthropic → Bedrock (upstream over HTTPS) | Cost window major page faults | 0 | Valid | — |
| Anthropic → Bedrock (upstream over HTTPS) | Added time to first event p50 (µs) | — | Untestable | — |
| Anthropic → Bedrock (upstream over HTTPS) | Added time to first event p99 (µs) | — | Untestable | — |
| Anthropic → Bedrock (upstream over HTTPS) | Added inter-frame gap p50 (µs) | — | Untestable | — |
| Anthropic → Bedrock (upstream over HTTPS) | Added inter-frame gap p99 (µs) | — | Untestable | — |
| Anthropic → Bedrock (upstream over HTTPS) | Streams sustained (frames/s) | — | Untestable | — |
| Anthropic → Bedrock (upstream over HTTPS) | Streams sustained | — | Untestable | — |
| Anthropic → Bedrock (upstream over HTTPS) | Idle RSS (MiB) | 160.8 | Valid | — |
| Anthropic → Bedrock (upstream over HTTPS) | Steady-state RSS (MiB) | 950.2 | Valid | — |
| Anthropic → Bedrock (upstream over HTTPS) | Recovered RSS (MiB) | 1,054 | Valid | — |
| Anthropic → Bedrock (upstream over HTTPS) | Peak RSS (MiB) | 1,191.1 | Valid | — |
| Anthropic → Bedrock (upstream over HTTPS) | Peak RSS high-water mark (MiB) | 1,138.2 | Valid | — |
| Anthropic → Bedrock (upstream over HTTPS) | Time to plateau (s) | — | Not measured | — |
| Anthropic → Bedrock (upstream over HTTPS) | Memory growth (MiB/min) | 99.69 | Valid | — |
| Anthropic → Bedrock (upstream over HTTPS) | Memory load duration (s) | 300 | Valid | — |
| Anthropic → Bedrock (upstream over HTTPS) | Memory shape under load | 1 | Valid | — |
| Anthropic → Bedrock (upstream over HTTPS) | Idle memory shape | — | Not measured | — |
| Anthropic → Bedrock (upstream over HTTPS) | Idle memory static | 1 | Valid | — |
| Anthropic → Bedrock (upstream over HTTPS) | Idle memory growth (MiB/min) | 0.353 | Valid | — |
| Anthropic → Bedrock (upstream over HTTPS) | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| Anthropic → Bedrock (upstream over HTTPS) | Throughput (req/s) | 3,426 | concurrency 4; n=20,565 | 5 ms |
| Anthropic → Bedrock (upstream over HTTPS) | Throughput (req/s) | 3,624 | concurrency 8; n=21,752 | 10 ms |
| Anthropic → Bedrock (upstream over HTTPS) | Throughput (req/s) | 3,706 | concurrency 32; n=22,241 | 50 ms |
| Anthropic → Bedrock (upstream over HTTPS) | Throughput (req/s) | 3,706 | concurrency 32; n=22,241 | 100 ms |
| Anthropic → Bedrock (upstream over HTTPS) | Throughput (req/s) | 3,706 | concurrency 32; n=22,241 | Unbounded |
| Gemini → Bedrock (upstream over HTTPS) | Added latency p50 (µs) | 732 | Valid · n=7,747 gateway / 257,445 direct | — |
| Gemini → Bedrock (upstream over HTTPS) | Added latency p99 (µs) | 1,096 | Valid · n=7,747 gateway / 257,445 direct | — |
| Gemini → Bedrock (upstream over HTTPS) | Gateway p99 at concurrency 1 (µs) | 1,125 | Valid · n=7,747 | — |
| Gemini → Bedrock (upstream over HTTPS) | Direct-to-mock p99 at concurrency 1 (µs) | 29 | Valid · n=257,445 | — |
| Gemini → Bedrock (upstream over HTTPS) | CPU per request (µs) | 1,022.9 | Valid | — |
| Gemini → Bedrock (upstream over HTTPS) | Requests per CPU-second | 977.6 | Valid | — |
| Gemini → Bedrock (upstream over HTTPS) | Cost window concurrency | 8 | Valid | — |
| Gemini → Bedrock (upstream over HTTPS) | Cost window core utilisation | 0.986 | Valid | — |
| Gemini → Bedrock (upstream over HTTPS) | Cost window successful requests | 23,209 | Valid | — |
| Gemini → Bedrock (upstream over HTTPS) | Cost window throughput (req/s) | 3,867 | Valid | — |
| Gemini → Bedrock (upstream over HTTPS) | Cost window threads | 13 | Valid | — |
| Gemini → Bedrock (upstream over HTTPS) | Non-voluntary context switches per request | 0.019 | Valid | — |
| Gemini → Bedrock (upstream over HTTPS) | Cost window major page faults | 0 | Valid | — |
| Gemini → Bedrock (upstream over HTTPS) | Added time to first event p50 (µs) | — | Untestable | — |
| Gemini → Bedrock (upstream over HTTPS) | Added time to first event p99 (µs) | — | Untestable | — |
| Gemini → Bedrock (upstream over HTTPS) | Added inter-frame gap p50 (µs) | — | Untestable | — |
| Gemini → Bedrock (upstream over HTTPS) | Added inter-frame gap p99 (µs) | — | Untestable | — |
| Gemini → Bedrock (upstream over HTTPS) | Streams sustained (frames/s) | — | Untestable | — |
| Gemini → Bedrock (upstream over HTTPS) | Streams sustained | — | Untestable | — |
| Gemini → Bedrock (upstream over HTTPS) | Idle RSS (MiB) | 159.6 | Valid | — |
| Gemini → Bedrock (upstream over HTTPS) | Steady-state RSS (MiB) | 951.7 | Valid | — |
| Gemini → Bedrock (upstream over HTTPS) | Recovered RSS (MiB) | 1,257.6 | Valid | — |
| Gemini → Bedrock (upstream over HTTPS) | Peak RSS (MiB) | 1,257.6 | Valid | — |
| Gemini → Bedrock (upstream over HTTPS) | Peak RSS high-water mark (MiB) | 1,150.5 | Valid | — |
| Gemini → Bedrock (upstream over HTTPS) | Time to plateau (s) | — | Not measured | — |
| Gemini → Bedrock (upstream over HTTPS) | Memory growth (MiB/min) | 113.8 | Valid | — |
| Gemini → Bedrock (upstream over HTTPS) | Memory load duration (s) | 300 | Valid | — |
| Gemini → Bedrock (upstream over HTTPS) | Memory shape under load | 1 | Valid | — |
| Gemini → Bedrock (upstream over HTTPS) | Idle memory shape | — | Not measured | — |
| Gemini → Bedrock (upstream over HTTPS) | Idle memory static | 1 | Valid | — |
| Gemini → Bedrock (upstream over HTTPS) | Idle memory growth (MiB/min) | 0.389 | Valid | — |
| Gemini → Bedrock (upstream over HTTPS) | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| Gemini → Bedrock (upstream over HTTPS) | Throughput (req/s) | 3,839 | concurrency 4; n=23,038 | 5 ms |
| Gemini → Bedrock (upstream over HTTPS) | Throughput (req/s) | 3,970 | concurrency 8; n=23,829 | 10 ms |
| Gemini → Bedrock (upstream over HTTPS) | Throughput (req/s) | 4,018 | concurrency 32; n=24,117 | 50 ms |
| Gemini → Bedrock (upstream over HTTPS) | Throughput (req/s) | 4,018 | concurrency 32; n=24,117 | 100 ms |
| Gemini → Bedrock (upstream over HTTPS) | Throughput (req/s) | 4,018 | concurrency 32; n=24,117 | Unbounded |
| Cohere → Bedrock (upstream over HTTPS) | Added latency p50 (µs) | 692 | Valid · n=8,190 gateway / 262,997 direct | — |
| Cohere → Bedrock (upstream over HTTPS) | Added latency p99 (µs) | 1,061 | Valid · n=8,190 gateway / 262,997 direct | — |
| Cohere → Bedrock (upstream over HTTPS) | Gateway p99 at concurrency 1 (µs) | 1,090 | Valid · n=8,190 | — |
| Cohere → Bedrock (upstream over HTTPS) | Direct-to-mock p99 at concurrency 1 (µs) | 29 | Valid · n=262,997 | — |
| Cohere → Bedrock (upstream over HTTPS) | CPU per request (µs) | 938.6 | Valid | — |
| Cohere → Bedrock (upstream over HTTPS) | Requests per CPU-second | 1,065.4 | Valid | — |
| Cohere → Bedrock (upstream over HTTPS) | Cost window concurrency | 8 | Valid | — |
| Cohere → Bedrock (upstream over HTTPS) | Cost window core utilisation | 0.981 | Valid | — |
| Cohere → Bedrock (upstream over HTTPS) | Cost window successful requests | 25,165 | Valid | — |
| Cohere → Bedrock (upstream over HTTPS) | Cost window throughput (req/s) | 4,193 | Valid | — |
| Cohere → Bedrock (upstream over HTTPS) | Cost window threads | 13 | Valid | — |
| Cohere → Bedrock (upstream over HTTPS) | Non-voluntary context switches per request | 0 | Valid | — |
| Cohere → Bedrock (upstream over HTTPS) | Cost window major page faults | 0 | Valid | — |
| Cohere → Bedrock (upstream over HTTPS) | Added time to first event p50 (µs) | — | Untestable | — |
| Cohere → Bedrock (upstream over HTTPS) | Added time to first event p99 (µs) | — | Untestable | — |
| Cohere → Bedrock (upstream over HTTPS) | Added inter-frame gap p50 (µs) | — | Untestable | — |
| Cohere → Bedrock (upstream over HTTPS) | Added inter-frame gap p99 (µs) | — | Untestable | — |
| Cohere → Bedrock (upstream over HTTPS) | Streams sustained (frames/s) | — | Untestable | — |
| Cohere → Bedrock (upstream over HTTPS) | Streams sustained | — | Untestable | — |
| Cohere → Bedrock (upstream over HTTPS) | Idle RSS (MiB) | 162.1 | Valid | — |
| Cohere → Bedrock (upstream over HTTPS) | Steady-state RSS (MiB) | 1,001.6 | Valid | — |
| Cohere → Bedrock (upstream over HTTPS) | Recovered RSS (MiB) | 1,041 | Valid | — |
| Cohere → Bedrock (upstream over HTTPS) | Peak RSS (MiB) | 1,135.6 | Valid | — |
| Cohere → Bedrock (upstream over HTTPS) | Peak RSS high-water mark (MiB) | 1,107.1 | Valid | — |
| Cohere → Bedrock (upstream over HTTPS) | Time to plateau (s) | — | Not measured | — |
| Cohere → Bedrock (upstream over HTTPS) | Memory growth (MiB/min) | 66.49 | Valid | — |
| Cohere → Bedrock (upstream over HTTPS) | Memory load duration (s) | 300 | Valid | — |
| Cohere → Bedrock (upstream over HTTPS) | Memory shape under load | 1 | Valid | — |
| Cohere → Bedrock (upstream over HTTPS) | Idle memory shape | — | Not measured | — |
| Cohere → Bedrock (upstream over HTTPS) | Idle memory static | 1 | Valid | — |
| Cohere → Bedrock (upstream over HTTPS) | Idle memory growth (MiB/min) | 0.799 | Valid | — |
| Cohere → Bedrock (upstream over HTTPS) | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| Cohere → Bedrock (upstream over HTTPS) | Throughput (req/s) | 4,119 | concurrency 4; n=24,719 | 5 ms |
| Cohere → Bedrock (upstream over HTTPS) | Throughput (req/s) | 4,263 | concurrency 8; n=25,581 | 10 ms |
| Cohere → Bedrock (upstream over HTTPS) | Throughput (req/s) | 4,275 | concurrency 32; n=25,658 | 50 ms |
| Cohere → Bedrock (upstream over HTTPS) | Throughput (req/s) | 4,275 | concurrency 32; n=25,658 | 100 ms |
| Cohere → Bedrock (upstream over HTTPS) | Throughput (req/s) | 4,275 | concurrency 32; n=25,658 | Unbounded |
| Bedrock → Bedrock (upstream over HTTPS) | Added latency p50 (µs) | 694 | Valid · n=8,133 gateway / 263,582 direct | — |
| Bedrock → Bedrock (upstream over HTTPS) | Added latency p99 (µs) | 1,052 | Valid · n=8,133 gateway / 263,582 direct | — |
| Bedrock → Bedrock (upstream over HTTPS) | Gateway p99 at concurrency 1 (µs) | 1,079 | Valid · n=8,133 | — |
| Bedrock → Bedrock (upstream over HTTPS) | Direct-to-mock p99 at concurrency 1 (µs) | 27 | Valid · n=263,582 | — |
| Bedrock → Bedrock (upstream over HTTPS) | CPU per request (µs) | 956.8 | Valid | — |
| Bedrock → Bedrock (upstream over HTTPS) | Requests per CPU-second | 1,045.2 | Valid | — |
| Bedrock → Bedrock (upstream over HTTPS) | Cost window concurrency | 8 | Valid | — |
| Bedrock → Bedrock (upstream over HTTPS) | Cost window core utilisation | 0.989 | Valid | — |
| Bedrock → Bedrock (upstream over HTTPS) | Cost window successful requests | 24,886 | Valid | — |
| Bedrock → Bedrock (upstream over HTTPS) | Cost window throughput (req/s) | 4,147 | Valid | — |
| Bedrock → Bedrock (upstream over HTTPS) | Cost window threads | 14 | Valid | — |
| Bedrock → Bedrock (upstream over HTTPS) | Non-voluntary context switches per request | 0.034 | Valid | — |
| Bedrock → Bedrock (upstream over HTTPS) | Cost window major page faults | 0 | Valid | — |
| Bedrock → Bedrock (upstream over HTTPS) | Added time to first event p50 (µs) | — | Untestable | — |
| Bedrock → Bedrock (upstream over HTTPS) | Added time to first event p99 (µs) | — | Untestable | — |
| Bedrock → Bedrock (upstream over HTTPS) | Added inter-frame gap p50 (µs) | — | Untestable | — |
| Bedrock → Bedrock (upstream over HTTPS) | Added inter-frame gap p99 (µs) | — | Untestable | — |
| Bedrock → Bedrock (upstream over HTTPS) | Streams sustained (frames/s) | — | Untestable | — |
| Bedrock → Bedrock (upstream over HTTPS) | Streams sustained | — | Untestable | — |
| Bedrock → Bedrock (upstream over HTTPS) | Idle RSS (MiB) | 160 | Valid | — |
| Bedrock → Bedrock (upstream over HTTPS) | Steady-state RSS (MiB) | 964.1 | Valid | — |
| Bedrock → Bedrock (upstream over HTTPS) | Recovered RSS (MiB) | 1,242.1 | Valid | — |
| Bedrock → Bedrock (upstream over HTTPS) | Peak RSS (MiB) | 1,242.1 | Valid | — |
| Bedrock → Bedrock (upstream over HTTPS) | Peak RSS high-water mark (MiB) | 1,119.4 | Valid | — |
| Bedrock → Bedrock (upstream over HTTPS) | Time to plateau (s) | — | Not measured | — |
| Bedrock → Bedrock (upstream over HTTPS) | Memory growth (MiB/min) | 61.4 | Valid | — |
| Bedrock → Bedrock (upstream over HTTPS) | Memory load duration (s) | 300 | Valid | — |
| Bedrock → Bedrock (upstream over HTTPS) | Memory shape under load | 1 | Valid | — |
| Bedrock → Bedrock (upstream over HTTPS) | Idle memory shape | — | Not measured | — |
| Bedrock → Bedrock (upstream over HTTPS) | Idle memory static | 1 | Valid | — |
| Bedrock → Bedrock (upstream over HTTPS) | Idle memory growth (MiB/min) | 0.783 | Valid | — |
| Bedrock → Bedrock (upstream over HTTPS) | Throughput (req/s) | — | Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 ms |
| Bedrock → Bedrock (upstream over HTTPS) | Throughput (req/s) | 3,832 | concurrency 4; n=22,999 | 5 ms |
| Bedrock → Bedrock (upstream over HTTPS) | Throughput (req/s) | 3,972 | concurrency 8; n=23,835 | 10 ms |
| Bedrock → Bedrock (upstream over HTTPS) | Throughput (req/s) | 4,035 | concurrency 32; n=24,214 | 50 ms |
| Bedrock → Bedrock (upstream over HTTPS) | Throughput (req/s) | 4,035 | concurrency 32; n=24,214 | 100 ms |
| Bedrock → Bedrock (upstream over HTTPS) | Throughput (req/s) | 4,035 | concurrency 32; n=24,214 | Unbounded |
Probe evidence and instrument provenance
Historical capability notes may reference an earlier product version. They are not a new probe or a current support claim. Full RSS samples remain in the immutable raw JSON.
| Ingress → upstream | Cell state | Probe HTTP status | Recorded reason | Recorded detail | Evidence scope | Response excerpt | Tested build |
|---|---|---|---|---|---|---|---|
| OpenAI → OpenAI | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| OpenAI Responses → OpenAI | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| Anthropic → OpenAI | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| Gemini → OpenAI | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| Cohere → OpenAI | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| Bedrock → OpenAI | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| OpenAI → OpenAI Responses | Not declared | — | recorded note 1 | recorded note 1 | recorded note 2 | — | maximhq/bifrost:v2.5.0 |
| OpenAI Responses → OpenAI Responses | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| Anthropic → OpenAI Responses | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| Gemini → OpenAI Responses | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| Cohere → OpenAI Responses | Not declared | — | recorded note 1 | recorded note 1 | recorded note 2 | — | maximhq/bifrost:v2.5.0 |
| Bedrock → OpenAI Responses | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| OpenAI → Anthropic | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| OpenAI Responses → Anthropic | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| Anthropic → Anthropic | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| Gemini → Anthropic | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| Cohere → Anthropic | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| Bedrock → Anthropic | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| OpenAI → Gemini | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| OpenAI Responses → Gemini | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| Anthropic → Gemini | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| Gemini → Gemini | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| Cohere → Gemini | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| Bedrock → Gemini | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| OpenAI → Cohere | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| OpenAI Responses → Cohere | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| Anthropic → Cohere | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| Gemini → Cohere | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| Cohere → Cohere | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| Bedrock → Cohere | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| OpenAI → Bedrock | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| OpenAI Responses → Bedrock | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| Anthropic → Bedrock | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| Gemini → Bedrock | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| Cohere → Bedrock | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
| Bedrock → Bedrock | Served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0 |
- Harness commit
d240badd966933fa19eec0606b2d1f3945f18ad2- Mock SHA-256
99d01f709d8d9c1f49055aa265d5eb7475264d6b513911fe317aee74a71190f8- Load generator SHA-256
not recorded- net.core.somaxconn
- 65,535
- net.ipv4.tcp_max_syn_backlog
- 65,535
Stream sweep diagnostics
Per streaming-sweep concurrency, as recorded by the engine: lanes by final HTTP status, and lanes that received no response.
| Ingress → upstream | Concurrency | Lanes by HTTP status | Lanes with no response |
|---|---|---|---|
| OpenAI → OpenAI | 1 | HTTP 200: 1 | 0 |
| OpenAI → OpenAI | 2 | HTTP 200: 2 | 0 |
| OpenAI → OpenAI | 4 | HTTP 200: 4 | 0 |
| OpenAI → OpenAI | 8 | HTTP 200: 8 | 0 |
| OpenAI → OpenAI | 16 | HTTP 200: 16 | 0 |
| OpenAI → OpenAI | 32 | HTTP 200: 32 | 0 |
| OpenAI → OpenAI | 64 | HTTP 200: 64 | 0 |
| OpenAI → OpenAI | 128 | HTTP 200: 128 | 0 |
| OpenAI → OpenAI | 256 | HTTP 200: 256 | 0 |
| OpenAI → OpenAI | 512 | HTTP 200: 512 | 0 |
| OpenAI → OpenAI | 1,024 | HTTP 200: 1,024 | 0 |
| OpenAI → OpenAI | 768 | HTTP 200: 768 | 0 |
| OpenAI → OpenAI | 640 | HTTP 200: 640 | 0 |
| OpenAI → OpenAI | 576 | HTTP 200: 576 | 0 |
| OpenAI → OpenAI | 608 | HTTP 200: 608 | 0 |
| OpenAI → OpenAI | 592 | HTTP 200: 592 | 0 |
| OpenAI → OpenAI | 584 | HTTP 200: 584 | 0 |
| OpenAI → OpenAI | 580 | HTTP 200: 580 | 0 |
| OpenAI → OpenAI | 582 | HTTP 200: 582 | 0 |
| OpenAI → OpenAI | 581 | HTTP 200: 581 | 0 |
| OpenAI → OpenAI | 580 | HTTP 200: 580 | 0 |
| OpenAI → OpenAI | 580 | HTTP 200: 580 | 0 |
| Anthropic → OpenAI | 1 | HTTP 200: 1 | 0 |
| Anthropic → OpenAI | 2 | HTTP 200: 2 | 0 |
| Anthropic → OpenAI | 4 | HTTP 200: 4 | 0 |
| Anthropic → OpenAI | 8 | HTTP 200: 8 | 0 |
| Anthropic → OpenAI | 16 | HTTP 200: 16 | 0 |
| Anthropic → OpenAI | 32 | HTTP 200: 32 | 0 |
| Anthropic → OpenAI | 64 | HTTP 200: 64 | 0 |
| Anthropic → OpenAI | 128 | HTTP 200: 128 | 0 |
| Anthropic → OpenAI | 256 | HTTP 200: 256 | 0 |
| Anthropic → OpenAI | 512 | HTTP 200: 512 | 0 |
| Anthropic → OpenAI | 1,024 | HTTP 200: 1,024 | 0 |
| Anthropic → OpenAI | 768 | HTTP 200: 768 | 0 |
| Anthropic → OpenAI | 640 | HTTP 200: 640 | 0 |
| Anthropic → OpenAI | 704 | HTTP 200: 704 | 0 |
| Anthropic → OpenAI | 672 | HTTP 200: 672 | 0 |
| Anthropic → OpenAI | 688 | HTTP 200: 688 | 0 |
| Anthropic → OpenAI | 680 | HTTP 200: 680 | 0 |
| Anthropic → OpenAI | 676 | HTTP 200: 676 | 0 |
| Anthropic → OpenAI | 678 | HTTP 200: 678 | 0 |
| Anthropic → OpenAI | 679 | HTTP 200: 679 | 0 |
| Anthropic → OpenAI | 678 | HTTP 200: 678 | 0 |
| Anthropic → OpenAI | 678 | HTTP 200: 678 | 0 |
| OpenAI → Anthropic | 1 | HTTP 200: 1 | 0 |
| OpenAI → Anthropic | 2 | HTTP 200: 2 | 0 |
| OpenAI → Anthropic | 4 | HTTP 200: 4 | 0 |
| OpenAI → Anthropic | 8 | HTTP 200: 8 | 0 |
| OpenAI → Anthropic | 16 | HTTP 200: 16 | 0 |
| OpenAI → Anthropic | 32 | HTTP 200: 32 | 0 |
| OpenAI → Anthropic | 64 | HTTP 200: 64 | 0 |
| OpenAI → Anthropic | 128 | HTTP 200: 128 | 0 |
| OpenAI → Anthropic | 256 | HTTP 200: 256 | 0 |
| OpenAI → Anthropic | 512 | HTTP 200: 512 | 0 |
| OpenAI → Anthropic | 1,024 | HTTP 200: 1,024 | 0 |
| OpenAI → Anthropic | 768 | HTTP 200: 768 | 0 |
| OpenAI → Anthropic | 640 | HTTP 200: 640 | 0 |
| OpenAI → Anthropic | 576 | HTTP 200: 576 | 0 |
| OpenAI → Anthropic | 608 | HTTP 200: 608 | 0 |
| OpenAI → Anthropic | 624 | HTTP 200: 624 | 0 |
| OpenAI → Anthropic | 632 | HTTP 200: 632 | 0 |
| OpenAI → Anthropic | 628 | HTTP 200: 628 | 0 |
| OpenAI → Anthropic | 626 | HTTP 200: 626 | 0 |
| OpenAI → Anthropic | 625 | HTTP 200: 625 | 0 |
| OpenAI → Anthropic | 624 | HTTP 200: 624 | 0 |
| OpenAI → Anthropic | 624 | HTTP 200: 624 | 0 |
| Anthropic → Anthropic | 1 | HTTP 200: 1 | 0 |
| Anthropic → Anthropic | 2 | HTTP 200: 2 | 0 |
| Anthropic → Anthropic | 4 | HTTP 200: 4 | 0 |
| Anthropic → Anthropic | 8 | HTTP 200: 8 | 0 |
| Anthropic → Anthropic | 16 | HTTP 200: 16 | 0 |
| Anthropic → Anthropic | 32 | HTTP 200: 32 | 0 |
| Anthropic → Anthropic | 64 | HTTP 200: 64 | 0 |
| Anthropic → Anthropic | 128 | HTTP 200: 128 | 0 |
| Anthropic → Anthropic | 256 | HTTP 200: 256 | 0 |
| Anthropic → Anthropic | 512 | HTTP 200: 512 | 0 |
| Anthropic → Anthropic | 1,024 | HTTP 200: 1,024 | 0 |
| Anthropic → Anthropic | 768 | HTTP 200: 768 | 0 |
| Anthropic → Anthropic | 896 | HTTP 200: 896 | 0 |
| Anthropic → Anthropic | 832 | HTTP 200: 832 | 0 |
| Anthropic → Anthropic | 864 | HTTP 200: 864 | 0 |
| Anthropic → Anthropic | 848 | HTTP 200: 848 | 0 |
| Anthropic → Anthropic | 856 | HTTP 200: 856 | 0 |
| Anthropic → Anthropic | 852 | HTTP 200: 852 | 0 |
| Anthropic → Anthropic | 854 | HTTP 200: 854 | 0 |
| Anthropic → Anthropic | 855 | HTTP 200: 855 | 0 |
| Anthropic → Anthropic | 855 | HTTP 200: 855 | 0 |
| Anthropic → Anthropic | 855 | HTTP 200: 855 | 0 |
| Anthropic → Anthropic | 683 | HTTP 200: 683 | 0 |
| Anthropic → Anthropic | 683 | HTTP 200: 683 | 0 |
| Anthropic → Anthropic | 683 | HTTP 200: 683 | 0 |
Recorded notes
Text recorded with the cells above. Each note is shown once; rows refer to it by number.
- Bifrost v2.5.0 source 69f4d56d18b9ba00ac5f9cf969b4501a882b9967. Every upstream except Bedrock uses the field's common model gpt-4o-mini; Bedrock uses a Claude model id (anthropic.claude-3-5-sonnet-20240620-v1:0) so Bifrost stays on the bedrock-runtime Converse surface rather than routing an OpenAI model family to its OpenAI-compatible Mantle surface (core/providers/bedrock/bedrock.go resolveSurface). OpenAI Chat and Cohere ingress into the Responses-only upstream are not enabled: Bifrost only converts them when the model catalog marks the model Responses-only (plugins/compat/main.go::markForConversion), and selecting a different model (o3-pro) to trigger that path also triggers the compat plugin's default parameter dropping (it removed max_completion_tokens in the preflight), which the benchmark's out-of-the-box rule excludes. The Bedrock upstream uses Bifrost's default HTTPS runtime host (bedrock-runtime.us-east-1.amazonaws.com), which the engine v1.2 catch-all DNS resolves to the mock; its six cells are probed and carry upstream_transport "tls".
- Historical capability research, not a fresh probe of the tested build. Referenced versions may differ from gateway_build; do not infer current capability.
Evidence for the current measurements
- run-bifrost-c253d4668350f1793f587661b425702d5448ca78d04ec9144589514b6064ebd7 · complete raw record including RSS series
Configuration recorded by the run
{
"framework": {
"pricing": {
"pricing_url": "file:///app/data/pricing.json",
"pricing_sync_interval": 86400,
"model_parameters_url": "file:///app/data/model-parameters.json"
}
},
"governance": {
"virtual_keys": [
{
"id": "bench",
"name": "benchmark",
"value": "sk-bf-benchmark-dummy",
"is_active": true,
"allow_all_providers": true
}
]
},
"providers": {
"anthropic": {
"keys": [
{
"value": "sk-dummy",
"models": [
"claude-3-5-sonnet-20241022"
],
"weight": 1
}
],
"network_config": {
"allow_private_network": true,
"base_url": "http://127.0.0.1:8000"
}
},
"gemini": {
"keys": [
{
"value": "sk-dummy",
"models": [
"gemini-1.5-pro"
],
"weight": 1
}
],
"network_config": {
"allow_private_network": true,
"base_url": "http://127.0.0.1:8000"
}
},
"cohere": {
"keys": [
{
"value": "sk-dummy",
"models": [
"command-r-plus"
],
"weight": 1
}
],
"network_config": {
"allow_private_network": true,
"base_url": "http://127.0.0.1:8000"
}
},
"bedrock": {
"keys": [
{
"value": "",
"models": [
"anthropic.claude-3-5-sonnet-20240620-v1:0"
],
"weight": 1,
"bedrock_key_config": {
"access_key": "AKIAMOCKACCESSKEY",
"secret_key": "mock-secret-access-key",
"region": "us-east-1"
}
}
],
"network_config": {
"allow_private_network": true
}
},
"bench-chat": {
"keys": [
{
"value": "sk-dummy",
"models": [
"gpt-4o-mini"
],
"weight": 1
}
],
"network_config": {
"allow_private_network": true,
"base_url": "http://127.0.0.1:8000"
},
"custom_provider_config": {
"base_provider_type": "openai",
"allowed_requests": {
"list_models": true,
"chat_completion": true,
"chat_completion_stream": true
}
}
},
"bench-responses": {
"keys": [
{
"value": "sk-dummy",
"models": [
"gpt-4o-mini"
],
"weight": 1
}
],
"network_config": {
"allow_private_network": true,
"base_url": "http://127.0.0.1:8000"
},
"custom_provider_config": {
"base_provider_type": "openai",
"allowed_requests": {
"list_models": true,
"responses": true,
"responses_stream": true
}
}
}
}
}
Compare a shared protocol cell
- Bifrost vs agentgateway
- Bifrost vs AISIX (api7)
- Bifrost vs APISIX
- Bifrost vs Busbar
- Bifrost vs GoModel
- Bifrost vs Helicone
- Bifrost vs Higress
- Bifrost vs Kong
- Bifrost vs LiteLLM · Python
- Bifrost vs LiteLLM · Rust
- Bifrost vs One-API
- Bifrost vs Plano
- Bifrost vs Portkey
- Bifrost vs TensorZero