# 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](https://onthebench.ai/gateways/method) .

Go · recorded build: maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78. Measured: 2026-11-11T03:17:59Z.

[JSON](https://onthebench.ai/api/v1/gateways/bifrost)  ·  [Markdown](https://onthebench.ai/gateways/bifrost.md)  ·  [CSV](https://onthebench.ai/gateways/data.csv?gateway=bifrost&compact=1)  ·  [Project source](https://github.com/maximhq/bifrost)  ·  [Interactive dashboard](https://onthebench.ai/gateways)

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_us | 658 | valid; gateway_c1_samples=8543; direct_c1_samples=269310 | — |
| openai → openai | added_latency_p99_us | 1057 | valid; gateway_c1_samples=8543; direct_c1_samples=269310 | — |
| openai → openai | gateway_c1_p99_us | 1084 | valid; gateway_c1_samples=8543 | — |
| openai → openai | direct_c1_p99_us | 27 | valid; direct_c1_samples=269310 | — |
| openai → openai | cpu_us_per_request | 865.1196519216824 | valid | — |
| openai → openai | rps_per_cpu_second | 1155.9094719195307 | valid | — |
| openai → openai | cost_window_conc | 8 | valid | — |
| openai → openai | cost_core_utilisation | 0.9909277682234482 | valid | — |
| openai → openai | cost_window_ok | 27580 | valid | — |
| openai → openai | cost_window_rps | 4595 | valid | — |
| openai → openai | cost_threads | 13 | valid | — |
| openai → openai | cost_nonvol_ctxt_per_request | 0.02708484408992023 | valid | — |
| openai → openai | cost_majflt | 0 | valid | — |
| openai → openai | added_ttft_p50_us | 932 | valid; ttft_gw_samples=100; ttft_direct_samples=100 | — |
| openai → openai | added_ttft_p99_us | 1417 | valid; ttft_gw_samples=100; ttft_direct_samples=100 | — |
| openai → openai | added_gap_p50_us | 3 | valid; gap_gw_samples=63; gap_direct_samples=63 | — |
| openai → openai | added_gap_p99_us | below_resolution | below_resolution; gap_gw_samples=63; gap_direct_samples=63 | — |
| openai → openai | streams_sustained_fps | 25840.89500725109 | valid | — |
| openai → openai | streams_sustained | 580 | valid | — |
| openai → openai | idle_rss_mib | 160.2734375 | valid | — |
| openai → openai | steady_state_rss_mib | 1049.515625 | valid | — |
| openai → openai | recovered_rss_mib | 1256.6328125 | valid | — |
| openai → openai | peak_rss_mib | 1257.0390625 | valid | — |
| openai → openai | peak_rss_hwm_mib | 1137.43359375 | valid | — |
| openai → openai | time_to_plateau_s | not_measured | not_measured | — |
| openai → openai | growth_rate_mib_per_min | 56.82213581346736 | valid | — |
| openai → openai | load_s | 300 | valid | — |
| openai → openai | shape | 1 | valid | — |
| openai → openai | idle_shape | not_measured | not_measured | — |
| openai → openai | idle_static | 1 | valid | — |
| openai → openai | idle_growth_rate_mib_per_min | 0.4802766176196459 | valid | — |
| openai → openai | throughput_rps | bound_not_met req/s | 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 |
| openai → openai | throughput_rps | 4346 req/s | concurrency 4; samples 26081 | 5 |
| openai → openai | throughput_rps | 4555 req/s | concurrency 8; samples 27340 | 10 |
| openai → openai | throughput_rps | 4624 req/s | concurrency 16; samples 27748 | 50 |
| openai → openai | throughput_rps | 4624 req/s | concurrency 16; samples 27748 | 100 |
| openai → openai | throughput_rps | 4624 req/s | concurrency 16; samples 27748 | unbounded |
| openai-responses → openai | added_latency_p50_us | 586 | valid; gateway_c1_samples=9572; direct_c1_samples=263867 | — |
| openai-responses → openai | added_latency_p99_us | 944 | valid; gateway_c1_samples=9572; direct_c1_samples=263867 | — |
| openai-responses → openai | gateway_c1_p99_us | 971 | valid; gateway_c1_samples=9572 | — |
| openai-responses → openai | direct_c1_p99_us | 27 | valid; direct_c1_samples=263867 | — |
| openai-responses → openai | cpu_us_per_request | 842.668921299232 | valid | — |
| openai-responses → openai | rps_per_cpu_second | 1186.705685618729 | valid | — |
| openai-responses → openai | cost_window_conc | 8 | valid | — |
| openai-responses → openai | cost_core_utilisation | 0.9933326877124895 | valid | — |
| openai-responses → openai | cost_window_ok | 28386 | valid | — |
| openai-responses → openai | cost_window_rps | 4730 | valid | — |
| openai-responses → openai | cost_threads | 13 | valid | — |
| openai-responses → openai | cost_nonvol_ctxt_per_request | 0.00542520961037131 | valid | — |
| openai-responses → openai | cost_majflt | 0 | valid | — |
| openai-responses → openai | added_ttft_p50_us | untestable | untestable | — |
| openai-responses → openai | added_ttft_p99_us | untestable | untestable | — |
| openai-responses → openai | added_gap_p50_us | untestable | untestable | — |
| openai-responses → openai | added_gap_p99_us | untestable | untestable | — |
| openai-responses → openai | streams_sustained_fps | untestable | untestable | — |
| openai-responses → openai | streams_sustained | untestable | untestable | — |
| openai-responses → openai | idle_rss_mib | 161.203125 | valid | — |
| openai-responses → openai | steady_state_rss_mib | 1095.1640625 | valid | — |
| openai-responses → openai | recovered_rss_mib | 1073.67578125 | valid | — |
| openai-responses → openai | peak_rss_mib | 1252.703125 | valid | — |
| openai-responses → openai | peak_rss_hwm_mib | 1167.71484375 | valid | — |
| openai-responses → openai | time_to_plateau_s | not_measured | not_measured | — |
| openai-responses → openai | growth_rate_mib_per_min | 73.1085228333916 | valid | — |
| openai-responses → openai | load_s | 300 | valid | — |
| openai-responses → openai | shape | 1 | valid | — |
| openai-responses → openai | idle_shape | not_measured | not_measured | — |
| openai-responses → openai | idle_static | 1 | valid | — |
| openai-responses → openai | idle_growth_rate_mib_per_min | 0.35490141220737104 | valid | — |
| openai-responses → openai | throughput_rps | bound_not_met req/s | 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 |
| openai-responses → openai | throughput_rps | 4373 req/s | concurrency 4; samples 26247 | 5 |
| openai-responses → openai | throughput_rps | 4576 req/s | concurrency 8; samples 27465 | 10 |
| openai-responses → openai | throughput_rps | 4629 req/s | concurrency 16; samples 27782 | 50 |
| openai-responses → openai | throughput_rps | 4629 req/s | concurrency 16; samples 27782 | 100 |
| openai-responses → openai | throughput_rps | 4629 req/s | concurrency 16; samples 27782 | unbounded |
| anthropic → openai | added_latency_p50_us | 597 | valid; gateway_c1_samples=9413; direct_c1_samples=262256 | — |
| anthropic → openai | added_latency_p99_us | 968 | valid; gateway_c1_samples=9413; direct_c1_samples=262256 | — |
| anthropic → openai | gateway_c1_p99_us | 996 | valid; gateway_c1_samples=9413 | — |
| anthropic → openai | direct_c1_p99_us | 28 | valid; direct_c1_samples=262256 | — |
| anthropic → openai | cpu_us_per_request | 863.6478806964202 | valid | — |
| anthropic → openai | rps_per_cpu_second | 1157.8792958927074 | valid | — |
| anthropic → openai | cost_window_conc | 8 | valid | — |
| anthropic → openai | cost_core_utilisation | 0.9909830628870169 | valid | — |
| anthropic → openai | cost_window_ok | 27627 | valid | — |
| anthropic → openai | cost_window_rps | 4603 | valid | — |
| anthropic → openai | cost_threads | 14 | valid | — |
| anthropic → openai | cost_nonvol_ctxt_per_request | 0 | valid | — |
| anthropic → openai | cost_majflt | 0 | valid | — |
| anthropic → openai | added_ttft_p50_us | 868 | valid; ttft_gw_samples=100; ttft_direct_samples=100 | — |
| anthropic → openai | added_ttft_p99_us | 1061 | valid; ttft_gw_samples=100; ttft_direct_samples=100 | — |
| anthropic → openai | added_gap_p50_us | 4 | valid; gap_gw_samples=63; gap_direct_samples=63 | — |
| anthropic → openai | added_gap_p99_us | 28 | valid; gap_gw_samples=63; gap_direct_samples=63 | — |
| anthropic → openai | streams_sustained_fps | 29329.081029339704 | valid | — |
| anthropic → openai | streams_sustained | 678 | valid | — |
| anthropic → openai | idle_rss_mib | 161.00390625 | valid | — |
| anthropic → openai | steady_state_rss_mib | 1099.4375 | valid | — |
| anthropic → openai | recovered_rss_mib | 1073.2734375 | valid | — |
| anthropic → openai | peak_rss_mib | 1300.45703125 | valid | — |
| anthropic → openai | peak_rss_hwm_mib | 1209.50390625 | valid | — |
| anthropic → openai | time_to_plateau_s | not_measured | not_measured | — |
| anthropic → openai | growth_rate_mib_per_min | 88.44650309718233 | valid | — |
| anthropic → openai | load_s | 300 | valid | — |
| anthropic → openai | shape | 1 | valid | — |
| anthropic → openai | idle_shape | not_measured | not_measured | — |
| anthropic → openai | idle_static | 1 | valid | — |
| anthropic → openai | idle_growth_rate_mib_per_min | 0.9297672905162758 | valid | — |
| anthropic → openai | throughput_rps | 1531 req/s | concurrency 1; samples 9192 | 1 |
| anthropic → openai | throughput_rps | 4343 req/s | concurrency 4; samples 26062 | 5 |
| anthropic → openai | throughput_rps | 4579 req/s | concurrency 8; samples 27481 | 10 |
| anthropic → openai | throughput_rps | 4623 req/s | concurrency 16; samples 27743 | 50 |
| anthropic → openai | throughput_rps | 4623 req/s | concurrency 16; samples 27743 | 100 |
| anthropic → openai | throughput_rps | 4623 req/s | concurrency 16; samples 27743 | unbounded |
| gemini → openai | added_latency_p50_us | 588 | valid; gateway_c1_samples=9465; direct_c1_samples=259786 | — |
| gemini → openai | added_latency_p99_us | 972 | valid; gateway_c1_samples=9465; direct_c1_samples=259786 | — |
| gemini → openai | gateway_c1_p99_us | 1000 | valid; gateway_c1_samples=9465 | — |
| gemini → openai | direct_c1_p99_us | 28 | valid; direct_c1_samples=259786 | — |
| gemini → openai | cpu_us_per_request | 846.2167253083129 | valid | — |
| gemini → openai | rps_per_cpu_second | 1181.7303653926922 | valid | — |
| gemini → openai | cost_window_conc | 8 | valid | — |
| gemini → openai | cost_core_utilisation | 0.9890099259774329 | valid | — |
| gemini → openai | cost_window_ok | 28137 | valid | — |
| gemini → openai | cost_window_rps | 4688 | valid | — |
| gemini → openai | cost_threads | 13 | valid | — |
| gemini → openai | cost_nonvol_ctxt_per_request | 0.021039911859828696 | valid | — |
| gemini → openai | cost_majflt | 0 | valid | — |
| gemini → openai | added_ttft_p50_us | untestable | untestable | — |
| gemini → openai | added_ttft_p99_us | untestable | untestable | — |
| gemini → openai | added_gap_p50_us | untestable | untestable | — |
| gemini → openai | added_gap_p99_us | untestable | untestable | — |
| gemini → openai | streams_sustained_fps | untestable | untestable | — |
| gemini → openai | streams_sustained | untestable | untestable | — |
| gemini → openai | idle_rss_mib | 162.01171875 | valid | — |
| gemini → openai | steady_state_rss_mib | 1091.18359375 | valid | — |
| gemini → openai | recovered_rss_mib | 1074.71875 | valid | — |
| gemini → openai | peak_rss_mib | 1249.7890625 | valid | — |
| gemini → openai | peak_rss_hwm_mib | 1190.69921875 | valid | — |
| gemini → openai | time_to_plateau_s | not_measured | not_measured | — |
| gemini → openai | growth_rate_mib_per_min | 81.07866820710633 | valid | — |
| gemini → openai | load_s | 300 | valid | — |
| gemini → openai | shape | 1 | valid | — |
| gemini → openai | idle_shape | not_measured | not_measured | — |
| gemini → openai | idle_static | 1 | valid | — |
| gemini → openai | idle_growth_rate_mib_per_min | 0.2632439098184473 | valid | — |
| gemini → openai | throughput_rps | 1547 req/s | concurrency 1; samples 9287 | 1 |
| gemini → openai | throughput_rps | 4450 req/s | concurrency 4; samples 26704 | 5 |
| gemini → openai | throughput_rps | 4646 req/s | concurrency 8; samples 27884 | 10 |
| gemini → openai | throughput_rps | 4724 req/s | concurrency 16; samples 28353 | 50 |
| gemini → openai | throughput_rps | 4724 req/s | concurrency 16; samples 28353 | 100 |
| gemini → openai | throughput_rps | 4724 req/s | concurrency 16; samples 28353 | unbounded |
| cohere → openai | added_latency_p50_us | 572 | valid; gateway_c1_samples=9654; direct_c1_samples=215609 | — |
| cohere → openai | added_latency_p99_us | 912 | valid; gateway_c1_samples=9654; direct_c1_samples=215609 | — |
| cohere → openai | gateway_c1_p99_us | 946 | valid; gateway_c1_samples=9654 | — |
| cohere → openai | direct_c1_p99_us | 34 | valid; direct_c1_samples=215609 | — |
| cohere → openai | cpu_us_per_request | 813.5755585206473 | valid | — |
| cohere → openai | rps_per_cpu_second | 1229.1421362489486 | valid | — |
| cohere → openai | cost_window_conc | 8 | valid | — |
| cohere → openai | cost_core_utilisation | 0.9878134274950701 | valid | — |
| cohere → openai | cost_window_ok | 29229 | valid | — |
| cohere → openai | cost_window_rps | 4870 | valid | — |
| cohere → openai | cost_threads | 13 | valid | — |
| cohere → openai | cost_nonvol_ctxt_per_request | 0.02292244004242362 | valid | — |
| cohere → openai | cost_majflt | 0 | valid | — |
| cohere → openai | added_ttft_p50_us | untestable | untestable | — |
| cohere → openai | added_ttft_p99_us | untestable | untestable | — |
| cohere → openai | added_gap_p50_us | untestable | untestable | — |
| cohere → openai | added_gap_p99_us | untestable | untestable | — |
| cohere → openai | streams_sustained_fps | untestable | untestable | — |
| cohere → openai | streams_sustained | untestable | untestable | — |
| cohere → openai | idle_rss_mib | 158.37890625 | valid | — |
| cohere → openai | steady_state_rss_mib | 1049.91796875 | valid | — |
| cohere → openai | recovered_rss_mib | 1259.31640625 | valid | — |
| cohere → openai | peak_rss_mib | 1259.328125 | valid | — |
| cohere → openai | peak_rss_hwm_mib | 1138.40234375 | valid | — |
| cohere → openai | time_to_plateau_s | not_measured | not_measured | — |
| cohere → openai | growth_rate_mib_per_min | 78.12778798490757 | valid | — |
| cohere → openai | load_s | 300 | valid | — |
| cohere → openai | shape | 1 | valid | — |
| cohere → openai | idle_shape | not_measured | not_measured | — |
| cohere → openai | idle_static | 1 | valid | — |
| cohere → openai | idle_growth_rate_mib_per_min | 0.9327641164400249 | valid | — |
| cohere → openai | throughput_rps | 1624 req/s | concurrency 1; samples 9748 | 1 |
| cohere → openai | throughput_rps | 4853 req/s | concurrency 8; samples 29122 | 5 |
| cohere → openai | throughput_rps | 4853 req/s | concurrency 8; samples 29122 | 10 |
| cohere → openai | throughput_rps | 4898 req/s | concurrency 16; samples 29394 | 50 |
| cohere → openai | throughput_rps | 4898 req/s | concurrency 16; samples 29394 | 100 |
| cohere → openai | throughput_rps | 4898 req/s | concurrency 16; samples 29394 | unbounded |
| bedrock → openai | added_latency_p50_us | 632 | valid; gateway_c1_samples=8716; direct_c1_samples=262374 | — |
| bedrock → openai | added_latency_p99_us | 1148 | valid; gateway_c1_samples=8716; direct_c1_samples=262374 | — |
| bedrock → openai | gateway_c1_p99_us | 1176 | valid; gateway_c1_samples=8716 | — |
| bedrock → openai | direct_c1_p99_us | 28 | valid; direct_c1_samples=262374 | — |
| bedrock → openai | cpu_us_per_request | 909.2650044046114 | valid | — |
| bedrock → openai | rps_per_cpu_second | 1099.7893850042124 | valid | — |
| bedrock → openai | cost_window_conc | 8 | valid | — |
| bedrock → openai | cost_core_utilisation | 0.9856819939844713 | valid | — |
| bedrock → openai | cost_window_ok | 26109 | valid | — |
| bedrock → openai | cost_window_rps | 4350 | valid | — |
| bedrock → openai | cost_threads | 13 | valid | — |
| bedrock → openai | cost_nonvol_ctxt_per_request | 0.0371902409130951 | valid | — |
| bedrock → openai | cost_majflt | 0 | valid | — |
| bedrock → openai | added_ttft_p50_us | untestable | untestable | — |
| bedrock → openai | added_ttft_p99_us | untestable | untestable | — |
| bedrock → openai | added_gap_p50_us | untestable | untestable | — |
| bedrock → openai | added_gap_p99_us | untestable | untestable | — |
| bedrock → openai | streams_sustained_fps | untestable | untestable | — |
| bedrock → openai | streams_sustained | untestable | untestable | — |
| bedrock → openai | idle_rss_mib | 160.41796875 | valid | — |
| bedrock → openai | steady_state_rss_mib | 1042.69921875 | valid | — |
| bedrock → openai | recovered_rss_mib | 1232.00390625 | valid | — |
| bedrock → openai | peak_rss_mib | 1234.2578125 | valid | — |
| bedrock → openai | peak_rss_hwm_mib | 1146.58203125 | valid | — |
| bedrock → openai | time_to_plateau_s | not_measured | not_measured | — |
| bedrock → openai | growth_rate_mib_per_min | 73.76935084154414 | valid | — |
| bedrock → openai | load_s | 300 | valid | — |
| bedrock → openai | shape | 1 | valid | — |
| bedrock → openai | idle_shape | not_measured | not_measured | — |
| bedrock → openai | idle_static | 1 | valid | — |
| bedrock → openai | idle_growth_rate_mib_per_min | 0.1681216604557927 | valid | — |
| bedrock → openai | throughput_rps | bound_not_met req/s | 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 |
| bedrock → openai | throughput_rps | 4096 req/s | concurrency 4; samples 24582 | 5 |
| bedrock → openai | throughput_rps | 4328 req/s | concurrency 8; samples 25973 | 10 |
| bedrock → openai | throughput_rps | 4389 req/s | concurrency 16; samples 26340 | 50 |
| bedrock → openai | throughput_rps | 4389 req/s | concurrency 16; samples 26340 | 100 |
| bedrock → openai | throughput_rps | 4389 req/s | concurrency 16; samples 26340 | unbounded |
| openai → openai-responses | — | — | not_declared: 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". | — |
| openai-responses → openai-responses | added_latency_p50_us | 631 | valid; gateway_c1_samples=8878; direct_c1_samples=266170 | — |
| openai-responses → openai-responses | added_latency_p99_us | 1001 | valid; gateway_c1_samples=8878; direct_c1_samples=266170 | — |
| openai-responses → openai-responses | gateway_c1_p99_us | 1029 | valid; gateway_c1_samples=8878 | — |
| openai-responses → openai-responses | direct_c1_p99_us | 28 | valid; direct_c1_samples=266170 | — |
| openai-responses → openai-responses | cpu_us_per_request | 882.23316426641 | valid | — |
| openai-responses → openai-responses | rps_per_cpu_second | 1133.487201007134 | valid | — |
| openai-responses → openai-responses | cost_window_conc | 8 | valid | — |
| openai-responses → openai-responses | cost_core_utilisation | 0.9899135051863024 | valid | — |
| openai-responses → openai-responses | cost_window_ok | 27011 | valid | — |
| openai-responses → openai-responses | cost_window_rps | 4501 | valid | — |
| openai-responses → openai-responses | cost_threads | 15 | valid | — |
| openai-responses → openai-responses | cost_nonvol_ctxt_per_request | 0.019510569767872347 | valid | — |
| openai-responses → openai-responses | cost_majflt | 0 | valid | — |
| openai-responses → openai-responses | added_ttft_p50_us | untestable | untestable | — |
| openai-responses → openai-responses | added_ttft_p99_us | untestable | untestable | — |
| openai-responses → openai-responses | added_gap_p50_us | untestable | untestable | — |
| openai-responses → openai-responses | added_gap_p99_us | untestable | untestable | — |
| openai-responses → openai-responses | streams_sustained_fps | untestable | untestable | — |
| openai-responses → openai-responses | streams_sustained | untestable | untestable | — |
| openai-responses → openai-responses | idle_rss_mib | 160.6015625 | valid | — |
| openai-responses → openai-responses | steady_state_rss_mib | 1093.65625 | valid | — |
| openai-responses → openai-responses | recovered_rss_mib | 1288.046875 | valid | — |
| openai-responses → openai-responses | peak_rss_mib | 1288.0703125 | valid | — |
| openai-responses → openai-responses | peak_rss_hwm_mib | 1178.140625 | valid | — |
| openai-responses → openai-responses | time_to_plateau_s | not_measured | not_measured | — |
| openai-responses → openai-responses | growth_rate_mib_per_min | 64.4174808472209 | valid | — |
| openai-responses → openai-responses | load_s | 300 | valid | — |
| openai-responses → openai-responses | shape | 1 | valid | — |
| openai-responses → openai-responses | idle_shape | not_measured | not_measured | — |
| openai-responses → openai-responses | idle_static | 1 | valid | — |
| openai-responses → openai-responses | idle_growth_rate_mib_per_min | 0.22443919578251986 | valid | — |
| openai-responses → openai-responses | throughput_rps | bound_not_met req/s | 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 |
| openai-responses → openai-responses | throughput_rps | 4320 req/s | concurrency 4; samples 25924 | 5 |
| openai-responses → openai-responses | throughput_rps | 4443 req/s | concurrency 8; samples 26666 | 10 |
| openai-responses → openai-responses | throughput_rps | 4471 req/s | concurrency 16; samples 26831 | 50 |
| openai-responses → openai-responses | throughput_rps | 4471 req/s | concurrency 16; samples 26831 | 100 |
| openai-responses → openai-responses | throughput_rps | 4471 req/s | concurrency 16; samples 26831 | unbounded |
| anthropic → openai-responses | added_latency_p50_us | 641 | valid; gateway_c1_samples=8756; direct_c1_samples=262163 | — |
| anthropic → openai-responses | added_latency_p99_us | 1011 | valid; gateway_c1_samples=8756; direct_c1_samples=262163 | — |
| anthropic → openai-responses | gateway_c1_p99_us | 1038 | valid; gateway_c1_samples=8756 | — |
| anthropic → openai-responses | direct_c1_p99_us | 27 | valid; direct_c1_samples=262163 | — |
| anthropic → openai-responses | cpu_us_per_request | 886.3399374348279 | valid | — |
| anthropic → openai-responses | rps_per_cpu_second | 1128.235294117647 | valid | — |
| anthropic → openai-responses | cost_window_conc | 8 | valid | — |
| anthropic → openai-responses | cost_core_utilisation | 0.9886562313460872 | valid | — |
| anthropic → openai-responses | cost_window_ok | 26852 | valid | — |
| anthropic → openai-responses | cost_window_rps | 4474 | valid | — |
| anthropic → openai-responses | cost_threads | 13 | valid | — |
| anthropic → openai-responses | cost_nonvol_ctxt_per_request | 0.02267987486965589 | valid | — |
| anthropic → openai-responses | cost_majflt | 0 | valid | — |
| anthropic → openai-responses | added_ttft_p50_us | untestable | untestable | — |
| anthropic → openai-responses | added_ttft_p99_us | untestable | untestable | — |
| anthropic → openai-responses | added_gap_p50_us | untestable | untestable | — |
| anthropic → openai-responses | added_gap_p99_us | untestable | untestable | — |
| anthropic → openai-responses | streams_sustained_fps | untestable | untestable | — |
| anthropic → openai-responses | streams_sustained | untestable | untestable | — |
| anthropic → openai-responses | idle_rss_mib | 164.2421875 | valid | — |
| anthropic → openai-responses | steady_state_rss_mib | 1105.4921875 | valid | — |
| anthropic → openai-responses | recovered_rss_mib | 1137.90234375 | valid | — |
| anthropic → openai-responses | peak_rss_mib | 1237.328125 | valid | — |
| anthropic → openai-responses | peak_rss_hwm_mib | 1209.71484375 | valid | — |
| anthropic → openai-responses | time_to_plateau_s | not_measured | not_measured | — |
| anthropic → openai-responses | growth_rate_mib_per_min | 83.21399179381376 | valid | — |
| anthropic → openai-responses | load_s | 300 | valid | — |
| anthropic → openai-responses | shape | 1 | valid | — |
| anthropic → openai-responses | idle_shape | not_measured | not_measured | — |
| anthropic → openai-responses | idle_static | 1 | valid | — |
| anthropic → openai-responses | idle_growth_rate_mib_per_min | 0.9698354595007855 | valid | — |
| anthropic → openai-responses | throughput_rps | bound_not_met req/s | 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 |
| anthropic → openai-responses | throughput_rps | 4161 req/s | concurrency 4; samples 24972 | 5 |
| anthropic → openai-responses | throughput_rps | 4330 req/s | concurrency 8; samples 25985 | 10 |
| anthropic → openai-responses | throughput_rps | 4391 req/s | concurrency 16; samples 26357 | 50 |
| anthropic → openai-responses | throughput_rps | 4391 req/s | concurrency 16; samples 26357 | 100 |
| anthropic → openai-responses | throughput_rps | 4391 req/s | concurrency 16; samples 26357 | unbounded |
| gemini → openai-responses | added_latency_p50_us | 653 | valid; gateway_c1_samples=8613; direct_c1_samples=269135 | — |
| gemini → openai-responses | added_latency_p99_us | 1008 | valid; gateway_c1_samples=8613; direct_c1_samples=269135 | — |
| gemini → openai-responses | gateway_c1_p99_us | 1035 | valid; gateway_c1_samples=8613 | — |
| gemini → openai-responses | direct_c1_p99_us | 27 | valid; direct_c1_samples=269135 | — |
| gemini → openai-responses | cpu_us_per_request | 901.3135480940304 | valid | — |
| gemini → openai-responses | rps_per_cpu_second | 1109.4918101637968 | valid | — |
| gemini → openai-responses | cost_window_conc | 8 | valid | — |
| gemini → openai-responses | cost_core_utilisation | 0.9888138381760533 | valid | — |
| gemini → openai-responses | cost_window_ok | 26417 | valid | — |
| gemini → openai-responses | cost_window_rps | 4402 | valid | — |
| gemini → openai-responses | cost_threads | 14 | valid | — |
| gemini → openai-responses | cost_nonvol_ctxt_per_request | 0 | valid | — |
| gemini → openai-responses | cost_majflt | 0 | valid | — |
| gemini → openai-responses | added_ttft_p50_us | untestable | untestable | — |
| gemini → openai-responses | added_ttft_p99_us | untestable | untestable | — |
| gemini → openai-responses | added_gap_p50_us | untestable | untestable | — |
| gemini → openai-responses | added_gap_p99_us | untestable | untestable | — |
| gemini → openai-responses | streams_sustained_fps | untestable | untestable | — |
| gemini → openai-responses | streams_sustained | untestable | untestable | — |
| gemini → openai-responses | idle_rss_mib | 159.6796875 | valid | — |
| gemini → openai-responses | steady_state_rss_mib | 1083.3671875 | valid | — |
| gemini → openai-responses | recovered_rss_mib | 1239.97265625 | valid | — |
| gemini → openai-responses | peak_rss_mib | 1240.00390625 | valid | — |
| gemini → openai-responses | peak_rss_hwm_mib | 1183.16015625 | valid | — |
| gemini → openai-responses | time_to_plateau_s | not_measured | not_measured | — |
| gemini → openai-responses | growth_rate_mib_per_min | 66.32112791437604 | valid | — |
| gemini → openai-responses | load_s | 300 | valid | — |
| gemini → openai-responses | shape | 1 | valid | — |
| gemini → openai-responses | idle_shape | not_measured | not_measured | — |
| gemini → openai-responses | idle_static | 1 | valid | — |
| gemini → openai-responses | idle_growth_rate_mib_per_min | 0.5692884595172528 | valid | — |
| gemini → openai-responses | throughput_rps | bound_not_met req/s | 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 |
| gemini → openai-responses | throughput_rps | 4182 req/s | concurrency 4; samples 25101 | 5 |
| gemini → openai-responses | throughput_rps | 4323 req/s | concurrency 8; samples 25944 | 10 |
| gemini → openai-responses | throughput_rps | 4375 req/s | concurrency 16; samples 26254 | 50 |
| gemini → openai-responses | throughput_rps | 4375 req/s | concurrency 16; samples 26254 | 100 |
| gemini → openai-responses | throughput_rps | 4375 req/s | concurrency 16; samples 26254 | unbounded |
| cohere → openai-responses | — | — | not_declared: 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". | — |
| bedrock → openai-responses | added_latency_p50_us | 611 | valid; gateway_c1_samples=9156; direct_c1_samples=264307 | — |
| bedrock → openai-responses | added_latency_p99_us | 958 | valid; gateway_c1_samples=9156; direct_c1_samples=264307 | — |
| bedrock → openai-responses | gateway_c1_p99_us | 985 | valid; gateway_c1_samples=9156 | — |
| bedrock → openai-responses | direct_c1_p99_us | 27 | valid; direct_c1_samples=264307 | — |
| bedrock → openai-responses | cpu_us_per_request | 851.7911829641763 | valid | — |
| bedrock → openai-responses | rps_per_cpu_second | 1173.9966555183946 | valid | — |
| bedrock → openai-responses | cost_window_conc | 8 | valid | — |
| bedrock → openai-responses | cost_core_utilisation | 0.9932927434109152 | valid | — |
| bedrock → openai-responses | cost_window_ok | 28082 | valid | — |
| bedrock → openai-responses | cost_window_rps | 4679 | valid | — |
| bedrock → openai-responses | cost_threads | 14 | valid | — |
| bedrock → openai-responses | cost_nonvol_ctxt_per_request | 0.04376468912470622 | valid | — |
| bedrock → openai-responses | cost_majflt | 0 | valid | — |
| bedrock → openai-responses | added_ttft_p50_us | untestable | untestable | — |
| bedrock → openai-responses | added_ttft_p99_us | untestable | untestable | — |
| bedrock → openai-responses | added_gap_p50_us | untestable | untestable | — |
| bedrock → openai-responses | added_gap_p99_us | untestable | untestable | — |
| bedrock → openai-responses | streams_sustained_fps | untestable | untestable | — |
| bedrock → openai-responses | streams_sustained | untestable | untestable | — |
| bedrock → openai-responses | idle_rss_mib | 161.5546875 | valid | — |
| bedrock → openai-responses | steady_state_rss_mib | 1083.189453125 | valid | — |
| bedrock → openai-responses | recovered_rss_mib | 1295.55859375 | valid | — |
| bedrock → openai-responses | peak_rss_mib | 1295.55859375 | valid | — |
| bedrock → openai-responses | peak_rss_hwm_mib | 1184.015625 | valid | — |
| bedrock → openai-responses | time_to_plateau_s | not_measured | not_measured | — |
| bedrock → openai-responses | growth_rate_mib_per_min | 85.1344155013595 | valid | — |
| bedrock → openai-responses | load_s | 300 | valid | — |
| bedrock → openai-responses | shape | 1 | valid | — |
| bedrock → openai-responses | idle_shape | not_measured | not_measured | — |
| bedrock → openai-responses | idle_static | 1 | valid | — |
| bedrock → openai-responses | idle_growth_rate_mib_per_min | 0.5264685233519769 | valid | — |
| bedrock → openai-responses | throughput_rps | bound_not_met req/s | 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 |
| bedrock → openai-responses | throughput_rps | 4197 req/s | concurrency 4; samples 25188 | 5 |
| bedrock → openai-responses | throughput_rps | 4442 req/s | concurrency 8; samples 26658 | 10 |
| bedrock → openai-responses | throughput_rps | 4547 req/s | concurrency 32; samples 27285 | 50 |
| bedrock → openai-responses | throughput_rps | 4547 req/s | concurrency 32; samples 27285 | 100 |
| bedrock → openai-responses | throughput_rps | 4547 req/s | concurrency 32; samples 27285 | unbounded |
| openai → anthropic | added_latency_p50_us | 673 | valid; gateway_c1_samples=8297; direct_c1_samples=262017 | — |
| openai → anthropic | added_latency_p99_us | 1094 | valid; gateway_c1_samples=8297; direct_c1_samples=262017 | — |
| openai → anthropic | gateway_c1_p99_us | 1122 | valid; gateway_c1_samples=8297 | — |
| openai → anthropic | direct_c1_p99_us | 28 | valid; direct_c1_samples=262017 | — |
| openai → anthropic | cpu_us_per_request | 931.3495782567948 | valid | — |
| openai → anthropic | rps_per_cpu_second | 1073.7106918238992 | valid | — |
| openai → anthropic | cost_window_conc | 8 | valid | — |
| openai → anthropic | cost_core_utilisation | 0.9905722786350731 | valid | — |
| openai → anthropic | cost_window_ok | 25608 | valid | — |
| openai → anthropic | cost_window_rps | 4267 | valid | — |
| openai → anthropic | cost_threads | 13 | valid | — |
| openai → anthropic | cost_nonvol_ctxt_per_request | 0.021829115901280848 | valid | — |
| openai → anthropic | cost_majflt | 0 | valid | — |
| openai → anthropic | added_ttft_p50_us | 857 | valid; ttft_gw_samples=100; ttft_direct_samples=100 | — |
| openai → anthropic | added_ttft_p99_us | 1039 | valid; ttft_gw_samples=100; ttft_direct_samples=100 | — |
| openai → anthropic | added_gap_p50_us | 7 | valid; gap_gw_samples=63; gap_direct_samples=63 | — |
| openai → anthropic | added_gap_p99_us | 25 | valid; gap_gw_samples=63; gap_direct_samples=63 | — |
| openai → anthropic | streams_sustained_fps | 26538.00329954379 | valid | — |
| openai → anthropic | streams_sustained | 624 | valid | — |
| openai → anthropic | idle_rss_mib | 162.32421875 | valid | — |
| openai → anthropic | steady_state_rss_mib | 1067.142578125 | valid | — |
| openai → anthropic | recovered_rss_mib | 1319.26171875 | valid | — |
| openai → anthropic | peak_rss_mib | 1319.3046875 | valid | — |
| openai → anthropic | peak_rss_hwm_mib | 1199.890625 | valid | — |
| openai → anthropic | time_to_plateau_s | not_measured | not_measured | — |
| openai → anthropic | growth_rate_mib_per_min | 63.28789981255695 | valid | — |
| openai → anthropic | load_s | 300 | valid | — |
| openai → anthropic | shape | 1 | valid | — |
| openai → anthropic | idle_shape | not_measured | not_measured | — |
| openai → anthropic | idle_static | 1 | valid | — |
| openai → anthropic | idle_growth_rate_mib_per_min | 0.6335511770764554 | valid | — |
| openai → anthropic | throughput_rps | bound_not_met req/s | 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 |
| openai → anthropic | throughput_rps | 4001 req/s | concurrency 4; samples 24014 | 5 |
| openai → anthropic | throughput_rps | 4117 req/s | concurrency 8; samples 24710 | 10 |
| openai → anthropic | throughput_rps | 4177 req/s | concurrency 32; samples 25067 | 50 |
| openai → anthropic | throughput_rps | 4177 req/s | concurrency 32; samples 25067 | 100 |
| openai → anthropic | throughput_rps | 4177 req/s | concurrency 32; samples 25067 | unbounded |
| openai-responses → anthropic | added_latency_p50_us | 770 | valid; gateway_c1_samples=7332; direct_c1_samples=263386 | — |
| openai-responses → anthropic | added_latency_p99_us | 1259 | valid; gateway_c1_samples=7332; direct_c1_samples=263386 | — |
| openai-responses → anthropic | gateway_c1_p99_us | 1287 | valid; gateway_c1_samples=7332 | — |
| openai-responses → anthropic | direct_c1_p99_us | 28 | valid; direct_c1_samples=263386 | — |
| openai-responses → anthropic | cpu_us_per_request | 1057.1097647897363 | valid | — |
| openai-responses → anthropic | rps_per_cpu_second | 945.9755583649389 | valid | — |
| openai-responses → anthropic | cost_window_conc | 8 | valid | — |
| openai-responses → anthropic | cost_core_utilisation | 0.9855767917888869 | valid | — |
| openai-responses → anthropic | cost_window_ok | 22448 | valid | — |
| openai-responses → anthropic | cost_window_rps | 3740 | valid | — |
| openai-responses → anthropic | cost_threads | 13 | valid | — |
| openai-responses → anthropic | cost_nonvol_ctxt_per_request | 0 | valid | — |
| openai-responses → anthropic | cost_majflt | 0 | valid | — |
| openai-responses → anthropic | added_ttft_p50_us | untestable | untestable | — |
| openai-responses → anthropic | added_ttft_p99_us | untestable | untestable | — |
| openai-responses → anthropic | added_gap_p50_us | untestable | untestable | — |
| openai-responses → anthropic | added_gap_p99_us | untestable | untestable | — |
| openai-responses → anthropic | streams_sustained_fps | untestable | untestable | — |
| openai-responses → anthropic | streams_sustained | untestable | untestable | — |
| openai-responses → anthropic | idle_rss_mib | 159.08984375 | valid | — |
| openai-responses → anthropic | steady_state_rss_mib | 984.9296875 | valid | — |
| openai-responses → anthropic | recovered_rss_mib | 1050.9609375 | valid | — |
| openai-responses → anthropic | peak_rss_mib | 1220.609375 | valid | — |
| openai-responses → anthropic | peak_rss_hwm_mib | 1174.5546875 | valid | — |
| openai-responses → anthropic | time_to_plateau_s | not_measured | not_measured | — |
| openai-responses → anthropic | growth_rate_mib_per_min | 168.51907253239412 | valid | — |
| openai-responses → anthropic | load_s | 300 | valid | — |
| openai-responses → anthropic | shape | 1 | valid | — |
| openai-responses → anthropic | idle_shape | not_measured | not_measured | — |
| openai-responses → anthropic | idle_static | 1 | valid | — |
| openai-responses → anthropic | idle_growth_rate_mib_per_min | 0.4696816087972396 | valid | — |
| openai-responses → anthropic | throughput_rps | bound_not_met req/s | 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 |
| openai-responses → anthropic | throughput_rps | 3458 req/s | concurrency 4; samples 20760 | 5 |
| openai-responses → anthropic | throughput_rps | 3703 req/s | concurrency 8; samples 22224 | 10 |
| openai-responses → anthropic | throughput_rps | 3703 req/s | concurrency 8; samples 22224 | 50 |
| openai-responses → anthropic | throughput_rps | 3703 req/s | concurrency 8; samples 22224 | 100 |
| openai-responses → anthropic | throughput_rps | 3703 req/s | concurrency 8; samples 22224 | unbounded |
| anthropic → anthropic | added_latency_p50_us | 678 | valid; gateway_c1_samples=8283; direct_c1_samples=258757 | — |
| anthropic → anthropic | added_latency_p99_us | 1101 | valid; gateway_c1_samples=8283; direct_c1_samples=258757 | — |
| anthropic → anthropic | gateway_c1_p99_us | 1129 | valid; gateway_c1_samples=8283 | — |
| anthropic → anthropic | direct_c1_p99_us | 28 | valid; direct_c1_samples=258757 | — |
| anthropic → anthropic | cpu_us_per_request | 1001.9235594212595 | valid | — |
| anthropic → anthropic | rps_per_cpu_second | 998.0801335559265 | valid | — |
| anthropic → anthropic | cost_window_conc | 8 | valid | — |
| anthropic → anthropic | cost_core_utilisation | 0.9949352467472666 | valid | — |
| anthropic → anthropic | cost_window_ok | 23914 | valid | — |
| anthropic → anthropic | cost_window_rps | 3985 | valid | — |
| anthropic → anthropic | cost_threads | 13 | valid | — |
| anthropic → anthropic | cost_nonvol_ctxt_per_request | 0.036882161077193276 | valid | — |
| anthropic → anthropic | cost_majflt | 0 | valid | — |
| anthropic → anthropic | added_ttft_p50_us | 826 | valid; ttft_gw_samples=100; ttft_direct_samples=100 | — |
| anthropic → anthropic | added_ttft_p99_us | 924 | valid; ttft_gw_samples=100; ttft_direct_samples=100 | — |
| anthropic → anthropic | added_gap_p50_us | below_resolution | below_resolution; gap_gw_samples=63; gap_direct_samples=63 | — |
| anthropic → anthropic | added_gap_p99_us | below_resolution | below_resolution; gap_gw_samples=63; gap_direct_samples=63 | — |
| anthropic → anthropic | streams_sustained_fps | 28037.49790079561 | valid | — |
| anthropic → anthropic | streams_sustained | 683 | valid | — |
| anthropic → anthropic | idle_rss_mib | 161.85546875 | valid | — |
| anthropic → anthropic | steady_state_rss_mib | 1059.20703125 | valid | — |
| anthropic → anthropic | recovered_rss_mib | 1112.515625 | valid | — |
| anthropic → anthropic | peak_rss_mib | 1274.21875 | valid | — |
| anthropic → anthropic | peak_rss_hwm_mib | 1215.40625 | valid | — |
| anthropic → anthropic | time_to_plateau_s | not_measured | not_measured | — |
| anthropic → anthropic | growth_rate_mib_per_min | 80.73395904462717 | valid | — |
| anthropic → anthropic | load_s | 300 | valid | — |
| anthropic → anthropic | shape | 1 | valid | — |
| anthropic → anthropic | idle_shape | not_measured | not_measured | — |
| anthropic → anthropic | idle_static | 1 | valid | — |
| anthropic → anthropic | idle_growth_rate_mib_per_min | 0.7146363497807083 | valid | — |
| anthropic → anthropic | throughput_rps | bound_not_met req/s | 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 |
| anthropic → anthropic | throughput_rps | 3861 req/s | concurrency 4; samples 23174 | 5 |
| anthropic → anthropic | throughput_rps | 3988 req/s | concurrency 8; samples 23931 | 10 |
| anthropic → anthropic | throughput_rps | 4035 req/s | concurrency 16; samples 24216 | 50 |
| anthropic → anthropic | throughput_rps | 4035 req/s | concurrency 16; samples 24216 | 100 |
| anthropic → anthropic | throughput_rps | 4035 req/s | concurrency 16; samples 24216 | unbounded |
| gemini → anthropic | added_latency_p50_us | 693 | valid; gateway_c1_samples=8044; direct_c1_samples=259231 | — |
| gemini → anthropic | added_latency_p99_us | 1153 | valid; gateway_c1_samples=8044; direct_c1_samples=259231 | — |
| gemini → anthropic | gateway_c1_p99_us | 1181 | valid; gateway_c1_samples=8044 | — |
| gemini → anthropic | direct_c1_p99_us | 28 | valid; direct_c1_samples=259231 | — |
| gemini → anthropic | cpu_us_per_request | 998.9980796526676 | valid | — |
| gemini → anthropic | rps_per_cpu_second | 1001.0029251984956 | valid | — |
| gemini → anthropic | cost_window_conc | 8 | valid | — |
| gemini → anthropic | cost_core_utilisation | 0.9937373062060388 | valid | — |
| gemini → anthropic | cost_window_ok | 23954 | valid | — |
| gemini → anthropic | cost_window_rps | 3991 | valid | — |
| gemini → anthropic | cost_threads | 16 | valid | — |
| gemini → anthropic | cost_nonvol_ctxt_per_request | 0 | valid | — |
| gemini → anthropic | cost_majflt | 0 | valid | — |
| gemini → anthropic | added_ttft_p50_us | untestable | untestable | — |
| gemini → anthropic | added_ttft_p99_us | untestable | untestable | — |
| gemini → anthropic | added_gap_p50_us | untestable | untestable | — |
| gemini → anthropic | added_gap_p99_us | untestable | untestable | — |
| gemini → anthropic | streams_sustained_fps | untestable | untestable | — |
| gemini → anthropic | streams_sustained | untestable | untestable | — |
| gemini → anthropic | idle_rss_mib | 161.1328125 | valid | — |
| gemini → anthropic | steady_state_rss_mib | 1047.7734375 | valid | — |
| gemini → anthropic | recovered_rss_mib | 1109.0703125 | valid | — |
| gemini → anthropic | peak_rss_mib | 1249.7578125 | valid | — |
| gemini → anthropic | peak_rss_hwm_mib | 1215.25 | valid | — |
| gemini → anthropic | time_to_plateau_s | not_measured | not_measured | — |
| gemini → anthropic | growth_rate_mib_per_min | 96.97735724619463 | valid | — |
| gemini → anthropic | load_s | 300 | valid | — |
| gemini → anthropic | shape | 1 | valid | — |
| gemini → anthropic | idle_shape | not_measured | not_measured | — |
| gemini → anthropic | idle_static | 1 | valid | — |
| gemini → anthropic | idle_growth_rate_mib_per_min | 0.21219393604779083 | valid | — |
| gemini → anthropic | throughput_rps | bound_not_met req/s | 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 |
| gemini → anthropic | throughput_rps | 3872 req/s | concurrency 4; samples 23239 | 5 |
| gemini → anthropic | throughput_rps | 3995 req/s | concurrency 8; samples 23976 | 10 |
| gemini → anthropic | throughput_rps | 4040 req/s | concurrency 16; samples 24249 | 50 |
| gemini → anthropic | throughput_rps | 4040 req/s | concurrency 16; samples 24249 | 100 |
| gemini → anthropic | throughput_rps | 4040 req/s | concurrency 16; samples 24249 | unbounded |
| cohere → anthropic | added_latency_p50_us | 685 | valid; gateway_c1_samples=8100; direct_c1_samples=245712 | — |
| cohere → anthropic | added_latency_p99_us | 1201 | valid; gateway_c1_samples=8100; direct_c1_samples=245712 | — |
| cohere → anthropic | gateway_c1_p99_us | 1230 | valid; gateway_c1_samples=8100 | — |
| cohere → anthropic | direct_c1_p99_us | 29 | valid; direct_c1_samples=245712 | — |
| cohere → anthropic | cpu_us_per_request | 946.1870960066622 | valid | — |
| cohere → anthropic | rps_per_cpu_second | 1056.8734283319363 | valid | — |
| cohere → anthropic | cost_window_conc | 8 | valid | — |
| cohere → anthropic | cost_core_utilisation | 0.9910715187644553 | valid | — |
| cohere → anthropic | cost_window_ok | 25217 | valid | — |
| cohere → anthropic | cost_window_rps | 4202 | valid | — |
| cohere → anthropic | cost_threads | 13 | valid | — |
| cohere → anthropic | cost_nonvol_ctxt_per_request | 0.019708926517825277 | valid | — |
| cohere → anthropic | cost_majflt | 0 | valid | — |
| cohere → anthropic | added_ttft_p50_us | untestable | untestable | — |
| cohere → anthropic | added_ttft_p99_us | untestable | untestable | — |
| cohere → anthropic | added_gap_p50_us | untestable | untestable | — |
| cohere → anthropic | added_gap_p99_us | untestable | untestable | — |
| cohere → anthropic | streams_sustained_fps | untestable | untestable | — |
| cohere → anthropic | streams_sustained | untestable | untestable | — |
| cohere → anthropic | idle_rss_mib | 163.01953125 | valid | — |
| cohere → anthropic | steady_state_rss_mib | 1019.55859375 | valid | — |
| cohere → anthropic | recovered_rss_mib | 961.4296875 | valid | — |
| cohere → anthropic | peak_rss_mib | 1261.3125 | valid | — |
| cohere → anthropic | peak_rss_hwm_mib | 1158.99609375 | valid | — |
| cohere → anthropic | time_to_plateau_s | not_measured | not_measured | — |
| cohere → anthropic | growth_rate_mib_per_min | 69.14945491716075 | valid | — |
| cohere → anthropic | load_s | 300 | valid | — |
| cohere → anthropic | shape | 1 | valid | — |
| cohere → anthropic | idle_shape | not_measured | not_measured | — |
| cohere → anthropic | idle_static | 1 | valid | — |
| cohere → anthropic | idle_growth_rate_mib_per_min | 0.26834700549162155 | valid | — |
| cohere → anthropic | throughput_rps | bound_not_met req/s | 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 |
| cohere → anthropic | throughput_rps | 3987 req/s | concurrency 4; samples 23927 | 5 |
| cohere → anthropic | throughput_rps | 4149 req/s | concurrency 8; samples 24902 | 10 |
| cohere → anthropic | throughput_rps | 4224 req/s | concurrency 16; samples 25349 | 50 |
| cohere → anthropic | throughput_rps | 4224 req/s | concurrency 16; samples 25349 | 100 |
| cohere → anthropic | throughput_rps | 4224 req/s | concurrency 16; samples 25349 | unbounded |
| bedrock → anthropic | added_latency_p50_us | 620 | valid; gateway_c1_samples=8987; direct_c1_samples=265983 | — |
| bedrock → anthropic | added_latency_p99_us | 1030 | valid; gateway_c1_samples=8987; direct_c1_samples=265983 | — |
| bedrock → anthropic | gateway_c1_p99_us | 1057 | valid; gateway_c1_samples=8987 | — |
| bedrock → anthropic | direct_c1_p99_us | 27 | valid; direct_c1_samples=265983 | — |
| bedrock → anthropic | cpu_us_per_request | 869.4860306706007 | valid | — |
| bedrock → anthropic | rps_per_cpu_second | 1150.1047339757017 | valid | — |
| bedrock → anthropic | cost_window_conc | 8 | valid | — |
| bedrock → anthropic | cost_core_utilisation | 0.9910610498359808 | valid | — |
| bedrock → anthropic | cost_window_ok | 27453 | valid | — |
| bedrock → anthropic | cost_window_rps | 4574 | valid | — |
| bedrock → anthropic | cost_threads | 13 | valid | — |
| bedrock → anthropic | cost_nonvol_ctxt_per_request | 0 | valid | — |
| bedrock → anthropic | cost_majflt | 0 | valid | — |
| bedrock → anthropic | added_ttft_p50_us | untestable | untestable | — |
| bedrock → anthropic | added_ttft_p99_us | untestable | untestable | — |
| bedrock → anthropic | added_gap_p50_us | untestable | untestable | — |
| bedrock → anthropic | added_gap_p99_us | untestable | untestable | — |
| bedrock → anthropic | streams_sustained_fps | untestable | untestable | — |
| bedrock → anthropic | streams_sustained | untestable | untestable | — |
| bedrock → anthropic | idle_rss_mib | 160.27734375 | valid | — |
| bedrock → anthropic | steady_state_rss_mib | 1059.34375 | valid | — |
| bedrock → anthropic | recovered_rss_mib | 1283.984375 | valid | — |
| bedrock → anthropic | peak_rss_mib | 1283.98828125 | valid | — |
| bedrock → anthropic | peak_rss_hwm_mib | 1155.984375 | valid | — |
| bedrock → anthropic | time_to_plateau_s | not_measured | not_measured | — |
| bedrock → anthropic | growth_rate_mib_per_min | 44.08274820353458 | valid | — |
| bedrock → anthropic | load_s | 300 | valid | — |
| bedrock → anthropic | shape | 1 | valid | — |
| bedrock → anthropic | idle_shape | not_measured | not_measured | — |
| bedrock → anthropic | idle_static | 1 | valid | — |
| bedrock → anthropic | idle_growth_rate_mib_per_min | 0.5597976599390831 | valid | — |
| bedrock → anthropic | throughput_rps | bound_not_met req/s | 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 |
| bedrock → anthropic | throughput_rps | 4524 req/s | concurrency 8; samples 27150 | 5 |
| bedrock → anthropic | throughput_rps | 4524 req/s | concurrency 8; samples 27150 | 10 |
| bedrock → anthropic | throughput_rps | 4633 req/s | concurrency 16; samples 27807 | 50 |
| bedrock → anthropic | throughput_rps | 4633 req/s | concurrency 16; samples 27807 | 100 |
| bedrock → anthropic | throughput_rps | 4633 req/s | concurrency 16; samples 27807 | unbounded |
| openai → gemini | added_latency_p50_us | 626 | valid; gateway_c1_samples=8969; direct_c1_samples=262364 | — |
| openai → gemini | added_latency_p99_us | 1018 | valid; gateway_c1_samples=8969; direct_c1_samples=262364 | — |
| openai → gemini | gateway_c1_p99_us | 1046 | valid; gateway_c1_samples=8969 | — |
| openai → gemini | direct_c1_p99_us | 28 | valid; direct_c1_samples=262364 | — |
| openai → gemini | cpu_us_per_request | 913.8050241991242 | valid | — |
| openai → gemini | rps_per_cpu_second | 1094.3253467843633 | valid | — |
| openai → gemini | cost_window_conc | 8 | valid | — |
| openai → gemini | cost_core_utilisation | 0.9882617364917406 | valid | — |
| openai → gemini | cost_window_ok | 26034 | valid | — |
| openai → gemini | cost_window_rps | 4338 | valid | — |
| openai → gemini | cost_threads | 14 | valid | — |
| openai → gemini | cost_nonvol_ctxt_per_request | 0 | valid | — |
| openai → gemini | cost_majflt | 0 | valid | — |
| openai → gemini | added_ttft_p50_us | untestable | untestable | — |
| openai → gemini | added_ttft_p99_us | untestable | untestable | — |
| openai → gemini | added_gap_p50_us | untestable | untestable | — |
| openai → gemini | added_gap_p99_us | untestable | untestable | — |
| openai → gemini | streams_sustained_fps | untestable | untestable | — |
| openai → gemini | streams_sustained | untestable | untestable | — |
| openai → gemini | idle_rss_mib | 162.02734375 | valid | — |
| openai → gemini | steady_state_rss_mib | 1093.5859375 | valid | — |
| openai → gemini | recovered_rss_mib | 1193.640625 | valid | — |
| openai → gemini | peak_rss_mib | 1199.39453125 | valid | — |
| openai → gemini | peak_rss_hwm_mib | 1200.12890625 | valid | — |
| openai → gemini | time_to_plateau_s | not_measured | not_measured | — |
| openai → gemini | growth_rate_mib_per_min | 73.29932566996371 | valid | — |
| openai → gemini | load_s | 300 | valid | — |
| openai → gemini | shape | 1 | valid | — |
| openai → gemini | idle_shape | not_measured | not_measured | — |
| openai → gemini | idle_static | 1 | valid | — |
| openai → gemini | idle_growth_rate_mib_per_min | 0.380561021774131 | valid | — |
| openai → gemini | throughput_rps | bound_not_met req/s | 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 |
| openai → gemini | throughput_rps | 4251 req/s | concurrency 4; samples 25514 | 5 |
| openai → gemini | throughput_rps | 4415 req/s | concurrency 8; samples 26497 | 10 |
| openai → gemini | throughput_rps | 4467 req/s | concurrency 16; samples 26809 | 50 |
| openai → gemini | throughput_rps | 4467 req/s | concurrency 16; samples 26809 | 100 |
| openai → gemini | throughput_rps | 4467 req/s | concurrency 16; samples 26809 | unbounded |
| openai-responses → gemini | added_latency_p50_us | 667 | valid; gateway_c1_samples=8355; direct_c1_samples=244434 | — |
| openai-responses → gemini | added_latency_p99_us | 1139 | valid; gateway_c1_samples=8355; direct_c1_samples=244434 | — |
| openai-responses → gemini | gateway_c1_p99_us | 1169 | valid; gateway_c1_samples=8355 | — |
| openai-responses → gemini | direct_c1_p99_us | 30 | valid; direct_c1_samples=244434 | — |
| openai-responses → gemini | cpu_us_per_request | 958.5794062110803 | valid | — |
| openai-responses → gemini | rps_per_cpu_second | 1043.2103939647945 | valid | — |
| openai-responses → gemini | cost_window_conc | 8 | valid | — |
| openai-responses → gemini | cost_core_utilisation | 0.9910643330296622 | valid | — |
| openai-responses → gemini | cost_window_ok | 24891 | valid | — |
| openai-responses → gemini | cost_window_rps | 4147 | valid | — |
| openai-responses → gemini | cost_threads | 14 | valid | — |
| openai-responses → gemini | cost_nonvol_ctxt_per_request | 0 | valid | — |
| openai-responses → gemini | cost_majflt | 0 | valid | — |
| openai-responses → gemini | added_ttft_p50_us | untestable | untestable | — |
| openai-responses → gemini | added_ttft_p99_us | untestable | untestable | — |
| openai-responses → gemini | added_gap_p50_us | untestable | untestable | — |
| openai-responses → gemini | added_gap_p99_us | untestable | untestable | — |
| openai-responses → gemini | streams_sustained_fps | untestable | untestable | — |
| openai-responses → gemini | streams_sustained | untestable | untestable | — |
| openai-responses → gemini | idle_rss_mib | 161.02734375 | valid | — |
| openai-responses → gemini | steady_state_rss_mib | 1065.328125 | valid | — |
| openai-responses → gemini | recovered_rss_mib | 1112.6953125 | valid | — |
| openai-responses → gemini | peak_rss_mib | 1287.9140625 | valid | — |
| openai-responses → gemini | peak_rss_hwm_mib | 1203.3515625 | valid | — |
| openai-responses → gemini | time_to_plateau_s | not_measured | not_measured | — |
| openai-responses → gemini | growth_rate_mib_per_min | 57.62264253641225 | valid | — |
| openai-responses → gemini | load_s | 300 | valid | — |
| openai-responses → gemini | shape | 1 | valid | — |
| openai-responses → gemini | idle_shape | not_measured | not_measured | — |
| openai-responses → gemini | idle_static | 1 | valid | — |
| openai-responses → gemini | idle_growth_rate_mib_per_min | 0.16256131294155626 | valid | — |
| openai-responses → gemini | throughput_rps | bound_not_met req/s | 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 |
| openai-responses → gemini | throughput_rps | 4033 req/s | concurrency 4; samples 24211 | 5 |
| openai-responses → gemini | throughput_rps | 4174 req/s | concurrency 8; samples 25048 | 10 |
| openai-responses → gemini | throughput_rps | 4222 req/s | concurrency 16; samples 25335 | 50 |
| openai-responses → gemini | throughput_rps | 4222 req/s | concurrency 16; samples 25335 | 100 |
| openai-responses → gemini | throughput_rps | 4222 req/s | concurrency 16; samples 25335 | unbounded |
| anthropic → gemini | added_latency_p50_us | 634 | valid; gateway_c1_samples=8607; direct_c1_samples=257441 | — |
| anthropic → gemini | added_latency_p99_us | 1125 | valid; gateway_c1_samples=8607; direct_c1_samples=257441 | — |
| anthropic → gemini | gateway_c1_p99_us | 1153 | valid; gateway_c1_samples=8607 | — |
| anthropic → gemini | direct_c1_p99_us | 28 | valid; direct_c1_samples=257441 | — |
| anthropic → gemini | cpu_us_per_request | 926.2870514820593 | valid | — |
| anthropic → gemini | rps_per_cpu_second | 1079.578947368421 | valid | — |
| anthropic → gemini | cost_window_conc | 8 | valid | — |
| anthropic → gemini | cost_core_utilisation | 0.986098460060681 | valid | — |
| anthropic → gemini | cost_window_ok | 25640 | valid | — |
| anthropic → gemini | cost_window_rps | 4272 | valid | — |
| anthropic → gemini | cost_threads | 14 | valid | — |
| anthropic → gemini | cost_nonvol_ctxt_per_request | 0.014430577223088924 | valid | — |
| anthropic → gemini | cost_majflt | 0 | valid | — |
| anthropic → gemini | added_ttft_p50_us | untestable | untestable | — |
| anthropic → gemini | added_ttft_p99_us | untestable | untestable | — |
| anthropic → gemini | added_gap_p50_us | untestable | untestable | — |
| anthropic → gemini | added_gap_p99_us | untestable | untestable | — |
| anthropic → gemini | streams_sustained_fps | untestable | untestable | — |
| anthropic → gemini | streams_sustained | untestable | untestable | — |
| anthropic → gemini | idle_rss_mib | 160.82421875 | valid | — |
| anthropic → gemini | steady_state_rss_mib | 1083.51953125 | valid | — |
| anthropic → gemini | recovered_rss_mib | 1179.125 | valid | — |
| anthropic → gemini | peak_rss_mib | 1239.10546875 | valid | — |
| anthropic → gemini | peak_rss_hwm_mib | 1228.04296875 | valid | — |
| anthropic → gemini | time_to_plateau_s | not_measured | not_measured | — |
| anthropic → gemini | growth_rate_mib_per_min | 81.52955434849295 | valid | — |
| anthropic → gemini | load_s | 300 | valid | — |
| anthropic → gemini | shape | 1 | valid | — |
| anthropic → gemini | idle_shape | not_measured | not_measured | — |
| anthropic → gemini | idle_static | 1 | valid | — |
| anthropic → gemini | idle_growth_rate_mib_per_min | 0.7108664636303431 | valid | — |
| anthropic → gemini | throughput_rps | bound_not_met req/s | 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 |
| anthropic → gemini | throughput_rps | 4055 req/s | concurrency 4; samples 24334 | 5 |
| anthropic → gemini | throughput_rps | 4182 req/s | concurrency 8; samples 25101 | 10 |
| anthropic → gemini | throughput_rps | 4257 req/s | concurrency 16; samples 25547 | 50 |
| anthropic → gemini | throughput_rps | 4257 req/s | concurrency 16; samples 25547 | 100 |
| anthropic → gemini | throughput_rps | 4257 req/s | concurrency 16; samples 25547 | unbounded |
| gemini → gemini | added_latency_p50_us | 652 | valid; gateway_c1_samples=8589; direct_c1_samples=266124 | — |
| gemini → gemini | added_latency_p99_us | 1098 | valid; gateway_c1_samples=8589; direct_c1_samples=266124 | — |
| gemini → gemini | gateway_c1_p99_us | 1125 | valid; gateway_c1_samples=8589 | — |
| gemini → gemini | direct_c1_p99_us | 27 | valid; direct_c1_samples=266124 | — |
| gemini → gemini | cpu_us_per_request | 954.9274255156608 | valid | — |
| gemini → gemini | rps_per_cpu_second | 1047.2 | valid | — |
| gemini → gemini | cost_window_conc | 8 | valid | — |
| gemini → gemini | cost_core_utilisation | 0.9866296894664301 | valid | — |
| gemini → gemini | cost_window_ok | 24871 | valid | — |
| gemini → gemini | cost_window_rps | 4144 | valid | — |
| gemini → gemini | cost_threads | 14 | valid | — |
| gemini → gemini | cost_nonvol_ctxt_per_request | 0 | valid | — |
| gemini → gemini | cost_majflt | 0 | valid | — |
| gemini → gemini | added_ttft_p50_us | untestable | untestable | — |
| gemini → gemini | added_ttft_p99_us | untestable | untestable | — |
| gemini → gemini | added_gap_p50_us | untestable | untestable | — |
| gemini → gemini | added_gap_p99_us | untestable | untestable | — |
| gemini → gemini | streams_sustained_fps | untestable | untestable | — |
| gemini → gemini | streams_sustained | untestable | untestable | — |
| gemini → gemini | idle_rss_mib | 161.47265625 | valid | — |
| gemini → gemini | steady_state_rss_mib | 1066.12109375 | valid | — |
| gemini → gemini | recovered_rss_mib | 863.8125 | valid | — |
| gemini → gemini | peak_rss_mib | 1200.625 | valid | — |
| gemini → gemini | peak_rss_hwm_mib | 1200.6953125 | valid | — |
| gemini → gemini | time_to_plateau_s | not_measured | not_measured | — |
| gemini → gemini | growth_rate_mib_per_min | 86.4146376641646 | valid | — |
| gemini → gemini | load_s | 300 | valid | — |
| gemini → gemini | shape | 1 | valid | — |
| gemini → gemini | idle_shape | not_measured | not_measured | — |
| gemini → gemini | idle_static | 1 | valid | — |
| gemini → gemini | idle_growth_rate_mib_per_min | 0.6505053739690356 | valid | — |
| gemini → gemini | throughput_rps | bound_not_met req/s | 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 |
| gemini → gemini | throughput_rps | 3959 req/s | concurrency 4; samples 23762 | 5 |
| gemini → gemini | throughput_rps | 4109 req/s | concurrency 8; samples 24661 | 10 |
| gemini → gemini | throughput_rps | 4162 req/s | concurrency 16; samples 24977 | 50 |
| gemini → gemini | throughput_rps | 4162 req/s | concurrency 16; samples 24977 | 100 |
| gemini → gemini | throughput_rps | 4162 req/s | concurrency 16; samples 24977 | unbounded |
| cohere → gemini | added_latency_p50_us | 658 | valid; gateway_c1_samples=8497; direct_c1_samples=243412 | — |
| cohere → gemini | added_latency_p99_us | 1090 | valid; gateway_c1_samples=8497; direct_c1_samples=243412 | — |
| cohere → gemini | gateway_c1_p99_us | 1120 | valid; gateway_c1_samples=8497 | — |
| cohere → gemini | direct_c1_p99_us | 30 | valid; direct_c1_samples=243412 | — |
| cohere → gemini | cpu_us_per_request | 953.6970279580229 | valid | — |
| cohere → gemini | rps_per_cpu_second | 1048.5510289794204 | valid | — |
| cohere → gemini | cost_window_conc | 8 | valid | — |
| cohere → gemini | cost_core_utilisation | 0.9889482427961908 | valid | — |
| cohere → gemini | cost_window_ok | 24966 | valid | — |
| cohere → gemini | cost_window_rps | 4160 | valid | — |
| cohere → gemini | cost_threads | 15 | valid | — |
| cohere → gemini | cost_nonvol_ctxt_per_request | 0 | valid | — |
| cohere → gemini | cost_majflt | 0 | valid | — |
| cohere → gemini | added_ttft_p50_us | untestable | untestable | — |
| cohere → gemini | added_ttft_p99_us | untestable | untestable | — |
| cohere → gemini | added_gap_p50_us | untestable | untestable | — |
| cohere → gemini | added_gap_p99_us | untestable | untestable | — |
| cohere → gemini | streams_sustained_fps | untestable | untestable | — |
| cohere → gemini | streams_sustained | untestable | untestable | — |
| cohere → gemini | idle_rss_mib | 160.05859375 | valid | — |
| cohere → gemini | steady_state_rss_mib | 986.193359375 | valid | — |
| cohere → gemini | recovered_rss_mib | 1046.546875 | valid | — |
| cohere → gemini | peak_rss_mib | 1179.5234375 | valid | — |
| cohere → gemini | peak_rss_hwm_mib | 1142.328125 | valid | — |
| cohere → gemini | time_to_plateau_s | not_measured | not_measured | — |
| cohere → gemini | growth_rate_mib_per_min | 69.0800380979254 | valid | — |
| cohere → gemini | load_s | 300 | valid | — |
| cohere → gemini | shape | 1 | valid | — |
| cohere → gemini | idle_shape | not_measured | not_measured | — |
| cohere → gemini | idle_static | 1 | valid | — |
| cohere → gemini | idle_growth_rate_mib_per_min | 0.438873930132072 | valid | — |
| cohere → gemini | throughput_rps | bound_not_met req/s | 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 |
| cohere → gemini | throughput_rps | 4011 req/s | concurrency 4; samples 24072 | 5 |
| cohere → gemini | throughput_rps | 4196 req/s | concurrency 8; samples 25181 | 10 |
| cohere → gemini | throughput_rps | 4233 req/s | concurrency 16; samples 25400 | 50 |
| cohere → gemini | throughput_rps | 4233 req/s | concurrency 16; samples 25400 | 100 |
| cohere → gemini | throughput_rps | 4233 req/s | concurrency 16; samples 25400 | unbounded |
| bedrock → gemini | added_latency_p50_us | 607 | valid; gateway_c1_samples=9211; direct_c1_samples=248434 | — |
| bedrock → gemini | added_latency_p99_us | 1009 | valid; gateway_c1_samples=9211; direct_c1_samples=248434 | — |
| bedrock → gemini | gateway_c1_p99_us | 1038 | valid; gateway_c1_samples=9211 | — |
| bedrock → gemini | direct_c1_p99_us | 29 | valid; direct_c1_samples=248434 | — |
| bedrock → gemini | cpu_us_per_request | 898.0235966032915 | valid | — |
| bedrock → gemini | rps_per_cpu_second | 1113.5564853556486 | valid | — |
| bedrock → gemini | cost_window_conc | 8 | valid | — |
| bedrock → gemini | cost_core_utilisation | 0.9925984018721842 | valid | — |
| bedrock → gemini | cost_window_ok | 26614 | valid | — |
| bedrock → gemini | cost_window_rps | 4435 | valid | — |
| bedrock → gemini | cost_threads | 13 | valid | — |
| bedrock → gemini | cost_nonvol_ctxt_per_request | 0 | valid | — |
| bedrock → gemini | cost_majflt | 0 | valid | — |
| bedrock → gemini | added_ttft_p50_us | untestable | untestable | — |
| bedrock → gemini | added_ttft_p99_us | untestable | untestable | — |
| bedrock → gemini | added_gap_p50_us | untestable | untestable | — |
| bedrock → gemini | added_gap_p99_us | untestable | untestable | — |
| bedrock → gemini | streams_sustained_fps | untestable | untestable | — |
| bedrock → gemini | streams_sustained | untestable | untestable | — |
| bedrock → gemini | idle_rss_mib | 160.74609375 | valid | — |
| bedrock → gemini | steady_state_rss_mib | 1057.921875 | valid | — |
| bedrock → gemini | recovered_rss_mib | 928.47265625 | valid | — |
| bedrock → gemini | peak_rss_mib | 1270.5859375 | valid | — |
| bedrock → gemini | peak_rss_hwm_mib | 1157.69921875 | valid | — |
| bedrock → gemini | time_to_plateau_s | not_measured | not_measured | — |
| bedrock → gemini | growth_rate_mib_per_min | 64.94421763716046 | valid | — |
| bedrock → gemini | load_s | 300 | valid | — |
| bedrock → gemini | shape | 1 | valid | — |
| bedrock → gemini | idle_shape | not_measured | not_measured | — |
| bedrock → gemini | idle_static | 1 | valid | — |
| bedrock → gemini | idle_growth_rate_mib_per_min | 0.33456776025435664 | valid | — |
| bedrock → gemini | throughput_rps | 1551 req/s | concurrency 1; samples 9310 | 1 |
| bedrock → gemini | throughput_rps | 4311 req/s | concurrency 4; samples 25871 | 5 |
| bedrock → gemini | throughput_rps | 4413 req/s | concurrency 8; samples 26482 | 10 |
| bedrock → gemini | throughput_rps | 4603 req/s | concurrency 16; samples 27624 | 50 |
| bedrock → gemini | throughput_rps | 4603 req/s | concurrency 16; samples 27624 | 100 |
| bedrock → gemini | throughput_rps | 4603 req/s | concurrency 16; samples 27624 | unbounded |
| openai → cohere | added_latency_p50_us | 641 | valid; gateway_c1_samples=8677; direct_c1_samples=263654 | — |
| openai → cohere | added_latency_p99_us | 1090 | valid; gateway_c1_samples=8677; direct_c1_samples=263654 | — |
| openai → cohere | gateway_c1_p99_us | 1118 | valid; gateway_c1_samples=8677 | — |
| openai → cohere | direct_c1_p99_us | 28 | valid; direct_c1_samples=263654 | — |
| openai → cohere | cpu_us_per_request | 900.0679296550683 | valid | — |
| openai → cohere | rps_per_cpu_second | 1111.027253668763 | valid | — |
| openai → cohere | cost_window_conc | 8 | valid | — |
| openai → cohere | cost_core_utilisation | 0.9906532496425215 | valid | — |
| openai → cohere | cost_window_ok | 26498 | valid | — |
| openai → cohere | cost_window_rps | 4415 | valid | — |
| openai → cohere | cost_threads | 14 | valid | — |
| openai → cohere | cost_nonvol_ctxt_per_request | 0 | valid | — |
| openai → cohere | cost_majflt | 0 | valid | — |
| openai → cohere | added_ttft_p50_us | untestable | untestable | — |
| openai → cohere | added_ttft_p99_us | untestable | untestable | — |
| openai → cohere | added_gap_p50_us | untestable | untestable | — |
| openai → cohere | added_gap_p99_us | untestable | untestable | — |
| openai → cohere | streams_sustained_fps | untestable | untestable | — |
| openai → cohere | streams_sustained | untestable | untestable | — |
| openai → cohere | idle_rss_mib | 161.38671875 | valid | — |
| openai → cohere | steady_state_rss_mib | 1119.109375 | valid | — |
| openai → cohere | recovered_rss_mib | 1071.81640625 | valid | — |
| openai → cohere | peak_rss_mib | 1254.93359375 | valid | — |
| openai → cohere | peak_rss_hwm_mib | 1211.171875 | valid | — |
| openai → cohere | time_to_plateau_s | not_measured | not_measured | — |
| openai → cohere | growth_rate_mib_per_min | 74.02020919745377 | valid | — |
| openai → cohere | load_s | 300 | valid | — |
| openai → cohere | shape | 1 | valid | — |
| openai → cohere | idle_shape | not_measured | not_measured | — |
| openai → cohere | idle_static | 1 | valid | — |
| openai → cohere | idle_growth_rate_mib_per_min | 0.15608876339144756 | valid | — |
| openai → cohere | throughput_rps | bound_not_met req/s | 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 |
| openai → cohere | throughput_rps | 4074 req/s | concurrency 4; samples 24453 | 5 |
| openai → cohere | throughput_rps | 4339 req/s | concurrency 8; samples 26039 | 10 |
| openai → cohere | throughput_rps | 4507 req/s | concurrency 32; samples 27048 | 50 |
| openai → cohere | throughput_rps | 4507 req/s | concurrency 32; samples 27048 | 100 |
| openai → cohere | throughput_rps | 4507 req/s | concurrency 32; samples 27048 | unbounded |
| openai-responses → cohere | added_latency_p50_us | 630 | valid; gateway_c1_samples=8900; direct_c1_samples=251031 | — |
| openai-responses → cohere | added_latency_p99_us | 1032 | valid; gateway_c1_samples=8900; direct_c1_samples=251031 | — |
| openai-responses → cohere | gateway_c1_p99_us | 1061 | valid; gateway_c1_samples=8900 | — |
| openai-responses → cohere | direct_c1_p99_us | 29 | valid; direct_c1_samples=251031 | — |
| openai-responses → cohere | cpu_us_per_request | 907.9445145018916 | valid | — |
| openai-responses → cohere | rps_per_cpu_second | 1101.388888888889 | valid | — |
| openai-responses → cohere | cost_window_conc | 8 | valid | — |
| openai-responses → cohere | cost_core_utilisation | 0.9869553319558813 | valid | — |
| openai-responses → cohere | cost_window_ok | 26169 | valid | — |
| openai-responses → cohere | cost_window_rps | 4360 | valid | — |
| openai-responses → cohere | cost_threads | 14 | valid | — |
| openai-responses → cohere | cost_nonvol_ctxt_per_request | 0.03695211891933203 | valid | — |
| openai-responses → cohere | cost_majflt | 0 | valid | — |
| openai-responses → cohere | added_ttft_p50_us | untestable | untestable | — |
| openai-responses → cohere | added_ttft_p99_us | untestable | untestable | — |
| openai-responses → cohere | added_gap_p50_us | untestable | untestable | — |
| openai-responses → cohere | added_gap_p99_us | untestable | untestable | — |
| openai-responses → cohere | streams_sustained_fps | untestable | untestable | — |
| openai-responses → cohere | streams_sustained | untestable | untestable | — |
| openai-responses → cohere | idle_rss_mib | 161.4140625 | valid | — |
| openai-responses → cohere | steady_state_rss_mib | 1074.9375 | valid | — |
| openai-responses → cohere | recovered_rss_mib | 1037.11328125 | valid | — |
| openai-responses → cohere | peak_rss_mib | 1256.90234375 | valid | — |
| openai-responses → cohere | peak_rss_hwm_mib | 1184.60546875 | valid | — |
| openai-responses → cohere | time_to_plateau_s | not_measured | not_measured | — |
| openai-responses → cohere | growth_rate_mib_per_min | 79.7896650787927 | valid | — |
| openai-responses → cohere | load_s | 300 | valid | — |
| openai-responses → cohere | shape | 1 | valid | — |
| openai-responses → cohere | idle_shape | not_measured | not_measured | — |
| openai-responses → cohere | idle_static | 1 | valid | — |
| openai-responses → cohere | idle_growth_rate_mib_per_min | 0.5104323063992409 | valid | — |
| openai-responses → cohere | throughput_rps | bound_not_met req/s | 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 |
| openai-responses → cohere | throughput_rps | 4135 req/s | concurrency 4; samples 24816 | 5 |
| openai-responses → cohere | throughput_rps | 4290 req/s | concurrency 8; samples 25751 | 10 |
| openai-responses → cohere | throughput_rps | 4330 req/s | concurrency 16; samples 25988 | 50 |
| openai-responses → cohere | throughput_rps | 4330 req/s | concurrency 16; samples 25988 | 100 |
| openai-responses → cohere | throughput_rps | 4330 req/s | concurrency 16; samples 25988 | unbounded |
| anthropic → cohere | added_latency_p50_us | 608 | valid; gateway_c1_samples=9106; direct_c1_samples=260492 | — |
| anthropic → cohere | added_latency_p99_us | 1061 | valid; gateway_c1_samples=9106; direct_c1_samples=260492 | — |
| anthropic → cohere | gateway_c1_p99_us | 1089 | valid; gateway_c1_samples=9106 | — |
| anthropic → cohere | direct_c1_p99_us | 28 | valid; direct_c1_samples=260492 | — |
| anthropic → cohere | cpu_us_per_request | 892.6769092744197 | valid | — |
| anthropic → cohere | rps_per_cpu_second | 1120.2261306532664 | valid | — |
| anthropic → cohere | cost_window_conc | 8 | valid | — |
| anthropic → cohere | cost_core_utilisation | 0.9914790593990157 | valid | — |
| anthropic → cohere | cost_window_ok | 26751 | valid | — |
| anthropic → cohere | cost_window_rps | 4457 | valid | — |
| anthropic → cohere | cost_threads | 14 | valid | — |
| anthropic → cohere | cost_nonvol_ctxt_per_request | 0.0182796904788606 | valid | — |
| anthropic → cohere | cost_majflt | 0 | valid | — |
| anthropic → cohere | added_ttft_p50_us | untestable | untestable | — |
| anthropic → cohere | added_ttft_p99_us | untestable | untestable | — |
| anthropic → cohere | added_gap_p50_us | untestable | untestable | — |
| anthropic → cohere | added_gap_p99_us | untestable | untestable | — |
| anthropic → cohere | streams_sustained_fps | untestable | untestable | — |
| anthropic → cohere | streams_sustained | untestable | untestable | — |
| anthropic → cohere | idle_rss_mib | 161.25390625 | valid | — |
| anthropic → cohere | steady_state_rss_mib | 1118.10546875 | valid | — |
| anthropic → cohere | recovered_rss_mib | 1140.0546875 | valid | — |
| anthropic → cohere | peak_rss_mib | 1222.47265625 | valid | — |
| anthropic → cohere | peak_rss_hwm_mib | 1216.37109375 | valid | — |
| anthropic → cohere | time_to_plateau_s | not_measured | not_measured | — |
| anthropic → cohere | growth_rate_mib_per_min | 68.59097881341208 | valid | — |
| anthropic → cohere | load_s | 300 | valid | — |
| anthropic → cohere | shape | 1 | valid | — |
| anthropic → cohere | idle_shape | not_measured | not_measured | — |
| anthropic → cohere | idle_static | 1 | valid | — |
| anthropic → cohere | idle_growth_rate_mib_per_min | 0.16652196535801708 | valid | — |
| anthropic → cohere | throughput_rps | bound_not_met req/s | 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 |
| anthropic → cohere | throughput_rps | 4182 req/s | concurrency 4; samples 25097 | 5 |
| anthropic → cohere | throughput_rps | 4364 req/s | concurrency 8; samples 26188 | 10 |
| anthropic → cohere | throughput_rps | 4404 req/s | concurrency 16; samples 26429 | 50 |
| anthropic → cohere | throughput_rps | 4404 req/s | concurrency 16; samples 26429 | 100 |
| anthropic → cohere | throughput_rps | 4404 req/s | concurrency 16; samples 26429 | unbounded |
| gemini → cohere | added_latency_p50_us | 670 | valid; gateway_c1_samples=8377; direct_c1_samples=268876 | — |
| gemini → cohere | added_latency_p99_us | 1111 | valid; gateway_c1_samples=8377; direct_c1_samples=268876 | — |
| gemini → cohere | gateway_c1_p99_us | 1138 | valid; gateway_c1_samples=8377 | — |
| gemini → cohere | direct_c1_p99_us | 27 | valid; direct_c1_samples=268876 | — |
| gemini → cohere | cpu_us_per_request | 944.4290190789735 | valid | — |
| gemini → cohere | rps_per_cpu_second | 1058.8408231835365 | valid | — |
| gemini → cohere | cost_window_conc | 8 | valid | — |
| gemini → cohere | cost_core_utilisation | 0.9888520783721307 | valid | — |
| gemini → cohere | cost_window_ok | 25211 | valid | — |
| gemini → cohere | cost_window_rps | 4200 | valid | — |
| gemini → cohere | cost_threads | 14 | valid | — |
| gemini → cohere | cost_nonvol_ctxt_per_request | 0.04561500932132799 | valid | — |
| gemini → cohere | cost_majflt | 0 | valid | — |
| gemini → cohere | added_ttft_p50_us | untestable | untestable | — |
| gemini → cohere | added_ttft_p99_us | untestable | untestable | — |
| gemini → cohere | added_gap_p50_us | untestable | untestable | — |
| gemini → cohere | added_gap_p99_us | untestable | untestable | — |
| gemini → cohere | streams_sustained_fps | untestable | untestable | — |
| gemini → cohere | streams_sustained | untestable | untestable | — |
| gemini → cohere | idle_rss_mib | 162.3125 | valid | — |
| gemini → cohere | steady_state_rss_mib | 1062.96484375 | valid | — |
| gemini → cohere | recovered_rss_mib | 1089.31640625 | valid | — |
| gemini → cohere | peak_rss_mib | 1247.7734375 | valid | — |
| gemini → cohere | peak_rss_hwm_mib | 1208.5546875 | valid | — |
| gemini → cohere | time_to_plateau_s | not_measured | not_measured | — |
| gemini → cohere | growth_rate_mib_per_min | 89.15210308843142 | valid | — |
| gemini → cohere | load_s | 300 | valid | — |
| gemini → cohere | shape | 1 | valid | — |
| gemini → cohere | idle_shape | not_measured | not_measured | — |
| gemini → cohere | idle_static | 1 | valid | — |
| gemini → cohere | idle_growth_rate_mib_per_min | 0.28888128372515204 | valid | — |
| gemini → cohere | throughput_rps | bound_not_met req/s | 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 |
| gemini → cohere | throughput_rps | 4019 req/s | concurrency 4; samples 24118 | 5 |
| gemini → cohere | throughput_rps | 4163 req/s | concurrency 8; samples 24987 | 10 |
| gemini → cohere | throughput_rps | 4191 req/s | concurrency 16; samples 25155 | 50 |
| gemini → cohere | throughput_rps | 4191 req/s | concurrency 16; samples 25155 | 100 |
| gemini → cohere | throughput_rps | 4191 req/s | concurrency 16; samples 25155 | unbounded |
| cohere → cohere | added_latency_p50_us | 609 | valid; gateway_c1_samples=9094; direct_c1_samples=264951 | — |
| cohere → cohere | added_latency_p99_us | 1062 | valid; gateway_c1_samples=9094; direct_c1_samples=264951 | — |
| cohere → cohere | gateway_c1_p99_us | 1090 | valid; gateway_c1_samples=9094 | — |
| cohere → cohere | direct_c1_p99_us | 28 | valid; direct_c1_samples=264951 | — |
| cohere → cohere | cpu_us_per_request | 888.3598080428555 | valid | — |
| cohere → cohere | rps_per_cpu_second | 1125.6700167504189 | valid | — |
| cohere → cohere | cost_window_conc | 8 | valid | — |
| cohere → cohere | cost_core_utilisation | 0.9919159700788478 | valid | — |
| cohere → cohere | cost_window_ok | 26881 | valid | — |
| cohere → cohere | cost_window_rps | 4479 | valid | — |
| cohere → cohere | cost_threads | 14 | valid | — |
| cohere → cohere | cost_nonvol_ctxt_per_request | 0.01860049849335962 | valid | — |
| cohere → cohere | cost_majflt | 0 | valid | — |
| cohere → cohere | added_ttft_p50_us | untestable | untestable | — |
| cohere → cohere | added_ttft_p99_us | untestable | untestable | — |
| cohere → cohere | added_gap_p50_us | untestable | untestable | — |
| cohere → cohere | added_gap_p99_us | untestable | untestable | — |
| cohere → cohere | streams_sustained_fps | untestable | untestable | — |
| cohere → cohere | streams_sustained | untestable | untestable | — |
| cohere → cohere | idle_rss_mib | 160.7265625 | valid | — |
| cohere → cohere | steady_state_rss_mib | 1074.453125 | valid | — |
| cohere → cohere | recovered_rss_mib | 1027.63671875 | valid | — |
| cohere → cohere | peak_rss_mib | 1245.73046875 | valid | — |
| cohere → cohere | peak_rss_hwm_mib | 1158.9921875 | valid | — |
| cohere → cohere | time_to_plateau_s | not_measured | not_measured | — |
| cohere → cohere | growth_rate_mib_per_min | 76.39987441289486 | valid | — |
| cohere → cohere | load_s | 300 | valid | — |
| cohere → cohere | shape | 1 | valid | — |
| cohere → cohere | idle_shape | not_measured | not_measured | — |
| cohere → cohere | idle_static | 1 | valid | — |
| cohere → cohere | idle_growth_rate_mib_per_min | 0.6278166214623293 | valid | — |
| cohere → cohere | throughput_rps | bound_not_met req/s | 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 |
| cohere → cohere | throughput_rps | 4379 req/s | concurrency 4; samples 26284 | 5 |
| cohere → cohere | throughput_rps | 4498 req/s | concurrency 8; samples 26992 | 10 |
| cohere → cohere | throughput_rps | 4544 req/s | concurrency 16; samples 27267 | 50 |
| cohere → cohere | throughput_rps | 4544 req/s | concurrency 16; samples 27267 | 100 |
| cohere → cohere | throughput_rps | 4544 req/s | concurrency 16; samples 27267 | unbounded |
| bedrock → cohere | added_latency_p50_us | 553 | valid; gateway_c1_samples=10081; direct_c1_samples=263966 | — |
| bedrock → cohere | added_latency_p99_us | 938 | valid; gateway_c1_samples=10081; direct_c1_samples=263966 | — |
| bedrock → cohere | gateway_c1_p99_us | 966 | valid; gateway_c1_samples=10081 | — |
| bedrock → cohere | direct_c1_p99_us | 28 | valid; direct_c1_samples=263966 | — |
| bedrock → cohere | cpu_us_per_request | 808.8659583856625 | valid | — |
| bedrock → cohere | rps_per_cpu_second | 1236.29878304658 | valid | — |
| bedrock → cohere | cost_window_conc | 8 | valid | — |
| bedrock → cohere | cost_core_utilisation | 0.9897773338096403 | valid | — |
| bedrock → cohere | cost_window_ok | 29461 | valid | — |
| bedrock → cohere | cost_window_rps | 4909 | valid | — |
| bedrock → cohere | cost_threads | 14 | valid | — |
| bedrock → cohere | cost_nonvol_ctxt_per_request | 0.016089066902006042 | valid | — |
| bedrock → cohere | cost_majflt | 0 | valid | — |
| bedrock → cohere | added_ttft_p50_us | untestable | untestable | — |
| bedrock → cohere | added_ttft_p99_us | untestable | untestable | — |
| bedrock → cohere | added_gap_p50_us | untestable | untestable | — |
| bedrock → cohere | added_gap_p99_us | untestable | untestable | — |
| bedrock → cohere | streams_sustained_fps | untestable | untestable | — |
| bedrock → cohere | streams_sustained | untestable | untestable | — |
| bedrock → cohere | idle_rss_mib | 161.31640625 | valid | — |
| bedrock → cohere | steady_state_rss_mib | 1081.46484375 | valid | — |
| bedrock → cohere | recovered_rss_mib | 1023.96875 | valid | — |
| bedrock → cohere | peak_rss_mib | 1293.359375 | valid | — |
| bedrock → cohere | peak_rss_hwm_mib | 1176.1015625 | valid | — |
| bedrock → cohere | time_to_plateau_s | not_measured | not_measured | — |
| bedrock → cohere | growth_rate_mib_per_min | 70.10000993732086 | valid | — |
| bedrock → cohere | load_s | 300 | valid | — |
| bedrock → cohere | shape | 1 | valid | — |
| bedrock → cohere | idle_shape | not_measured | not_measured | — |
| bedrock → cohere | idle_static | 1 | valid | — |
| bedrock → cohere | idle_growth_rate_mib_per_min | 0.3348033064043597 | valid | — |
| bedrock → cohere | throughput_rps | 1683 req/s | concurrency 1; samples 10101 | 1 |
| bedrock → cohere | throughput_rps | 4732 req/s | concurrency 4; samples 28401 | 5 |
| bedrock → cohere | throughput_rps | 4965 req/s | concurrency 8; samples 29795 | 10 |
| bedrock → cohere | throughput_rps | 5048 req/s | concurrency 16; samples 30292 | 50 |
| bedrock → cohere | throughput_rps | 5048 req/s | concurrency 16; samples 30292 | 100 |
| bedrock → cohere | throughput_rps | 5048 req/s | concurrency 16; samples 30292 | unbounded |
| openai → bedrock (upstream over HTTPS) | added_latency_p50_us | 708 | valid; gateway_c1_samples=7893; direct_c1_samples=260423 | — |
| openai → bedrock (upstream over HTTPS) | added_latency_p99_us | 1098 | valid; gateway_c1_samples=7893; direct_c1_samples=260423 | — |
| openai → bedrock (upstream over HTTPS) | gateway_c1_p99_us | 1126 | valid; gateway_c1_samples=7893 | — |
| openai → bedrock (upstream over HTTPS) | direct_c1_p99_us | 28 | valid; direct_c1_samples=260423 | — |
| openai → bedrock (upstream over HTTPS) | cpu_us_per_request | 955.841027699719 | valid | — |
| openai → bedrock (upstream over HTTPS) | rps_per_cpu_second | 1046.1990760184797 | valid | — |
| openai → bedrock (upstream over HTTPS) | cost_window_conc | 8 | valid | — |
| openai → bedrock (upstream over HTTPS) | cost_core_utilisation | 0.9890819979689783 | valid | — |
| openai → bedrock (upstream over HTTPS) | cost_window_ok | 24910 | valid | — |
| openai → bedrock (upstream over HTTPS) | cost_window_rps | 4151 | valid | — |
| openai → bedrock (upstream over HTTPS) | cost_threads | 17 | valid | — |
| openai → bedrock (upstream over HTTPS) | cost_nonvol_ctxt_per_request | 0 | valid | — |
| openai → bedrock (upstream over HTTPS) | cost_majflt | 0 | valid | — |
| openai → bedrock (upstream over HTTPS) | added_ttft_p50_us | untestable | untestable | — |
| openai → bedrock (upstream over HTTPS) | added_ttft_p99_us | untestable | untestable | — |
| openai → bedrock (upstream over HTTPS) | added_gap_p50_us | untestable | untestable | — |
| openai → bedrock (upstream over HTTPS) | added_gap_p99_us | untestable | untestable | — |
| openai → bedrock (upstream over HTTPS) | streams_sustained_fps | untestable | untestable | — |
| openai → bedrock (upstream over HTTPS) | streams_sustained | untestable | untestable | — |
| openai → bedrock (upstream over HTTPS) | idle_rss_mib | 159.95703125 | valid | — |
| openai → bedrock (upstream over HTTPS) | steady_state_rss_mib | 981.90234375 | valid | — |
| openai → bedrock (upstream over HTTPS) | recovered_rss_mib | 1138.0234375 | valid | — |
| openai → bedrock (upstream over HTTPS) | peak_rss_mib | 1138.05859375 | valid | — |
| openai → bedrock (upstream over HTTPS) | peak_rss_hwm_mib | 1131.03125 | valid | — |
| openai → bedrock (upstream over HTTPS) | time_to_plateau_s | not_measured | not_measured | — |
| openai → bedrock (upstream over HTTPS) | growth_rate_mib_per_min | 92.38327794178917 | valid | — |
| openai → bedrock (upstream over HTTPS) | load_s | 300 | valid | — |
| openai → bedrock (upstream over HTTPS) | shape | 1 | valid | — |
| openai → bedrock (upstream over HTTPS) | idle_shape | not_measured | not_measured | — |
| openai → bedrock (upstream over HTTPS) | idle_static | 1 | valid | — |
| openai → bedrock (upstream over HTTPS) | idle_growth_rate_mib_per_min | 0.7463544167999727 | valid | — |
| openai → bedrock (upstream over HTTPS) | throughput_rps | bound_not_met req/s | 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 |
| openai → bedrock (upstream over HTTPS) | throughput_rps | 3880 req/s | concurrency 4; samples 23291 | 5 |
| openai → bedrock (upstream over HTTPS) | throughput_rps | 4031 req/s | concurrency 8; samples 24194 | 10 |
| openai → bedrock (upstream over HTTPS) | throughput_rps | 4078 req/s | concurrency 32; samples 24475 | 50 |
| openai → bedrock (upstream over HTTPS) | throughput_rps | 4078 req/s | concurrency 32; samples 24475 | 100 |
| openai → bedrock (upstream over HTTPS) | throughput_rps | 4078 req/s | concurrency 32; samples 24475 | unbounded |
| openai-responses → bedrock (upstream over HTTPS) | added_latency_p50_us | 728 | valid; gateway_c1_samples=7776; direct_c1_samples=260884 | — |
| openai-responses → bedrock (upstream over HTTPS) | added_latency_p99_us | 1097 | valid; gateway_c1_samples=7776; direct_c1_samples=260884 | — |
| openai-responses → bedrock (upstream over HTTPS) | gateway_c1_p99_us | 1125 | valid; gateway_c1_samples=7776 | — |
| openai-responses → bedrock (upstream over HTTPS) | direct_c1_p99_us | 28 | valid; direct_c1_samples=260884 | — |
| openai-responses → bedrock (upstream over HTTPS) | cpu_us_per_request | 995.4637096774194 | valid | — |
| openai-responses → bedrock (upstream over HTTPS) | rps_per_cpu_second | 1004.5569620253165 | valid | — |
| openai-responses → bedrock (upstream over HTTPS) | cost_window_conc | 8 | valid | — |
| openai-responses → bedrock (upstream over HTTPS) | cost_core_utilisation | 0.9842016677515363 | valid | — |
| openai-responses → bedrock (upstream over HTTPS) | cost_window_ok | 23808 | valid | — |
| openai-responses → bedrock (upstream over HTTPS) | cost_window_rps | 3967 | valid | — |
| openai-responses → bedrock (upstream over HTTPS) | cost_threads | 14 | valid | — |
| openai-responses → bedrock (upstream over HTTPS) | cost_nonvol_ctxt_per_request | 0 | valid | — |
| openai-responses → bedrock (upstream over HTTPS) | cost_majflt | 0 | valid | — |
| openai-responses → bedrock (upstream over HTTPS) | added_ttft_p50_us | untestable | untestable | — |
| openai-responses → bedrock (upstream over HTTPS) | added_ttft_p99_us | untestable | untestable | — |
| openai-responses → bedrock (upstream over HTTPS) | added_gap_p50_us | untestable | untestable | — |
| openai-responses → bedrock (upstream over HTTPS) | added_gap_p99_us | untestable | untestable | — |
| openai-responses → bedrock (upstream over HTTPS) | streams_sustained_fps | untestable | untestable | — |
| openai-responses → bedrock (upstream over HTTPS) | streams_sustained | untestable | untestable | — |
| openai-responses → bedrock (upstream over HTTPS) | idle_rss_mib | 161.05078125 | valid | — |
| openai-responses → bedrock (upstream over HTTPS) | steady_state_rss_mib | 922.30078125 | valid | — |
| openai-responses → bedrock (upstream over HTTPS) | recovered_rss_mib | 1032.97265625 | valid | — |
| openai-responses → bedrock (upstream over HTTPS) | peak_rss_mib | 1145.89453125 | valid | — |
| openai-responses → bedrock (upstream over HTTPS) | peak_rss_hwm_mib | 1102.09375 | valid | — |
| openai-responses → bedrock (upstream over HTTPS) | time_to_plateau_s | not_measured | not_measured | — |
| openai-responses → bedrock (upstream over HTTPS) | growth_rate_mib_per_min | 104.23628620670526 | valid | — |
| openai-responses → bedrock (upstream over HTTPS) | load_s | 300 | valid | — |
| openai-responses → bedrock (upstream over HTTPS) | shape | 1 | valid | — |
| openai-responses → bedrock (upstream over HTTPS) | idle_shape | not_measured | not_measured | — |
| openai-responses → bedrock (upstream over HTTPS) | idle_static | 1 | valid | — |
| openai-responses → bedrock (upstream over HTTPS) | idle_growth_rate_mib_per_min | 0.2653526376532094 | valid | — |
| openai-responses → bedrock (upstream over HTTPS) | throughput_rps | bound_not_met req/s | 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 |
| openai-responses → bedrock (upstream over HTTPS) | throughput_rps | 3740 req/s | concurrency 4; samples 22445 | 5 |
| openai-responses → bedrock (upstream over HTTPS) | throughput_rps | 3865 req/s | concurrency 8; samples 23201 | 10 |
| openai-responses → bedrock (upstream over HTTPS) | throughput_rps | 3918 req/s | concurrency 32; samples 23511 | 50 |
| openai-responses → bedrock (upstream over HTTPS) | throughput_rps | 3918 req/s | concurrency 32; samples 23511 | 100 |
| openai-responses → bedrock (upstream over HTTPS) | throughput_rps | 3918 req/s | concurrency 32; samples 23511 | unbounded |
| anthropic → bedrock (upstream over HTTPS) | added_latency_p50_us | 726 | valid; gateway_c1_samples=7777; direct_c1_samples=244991 | — |
| anthropic → bedrock (upstream over HTTPS) | added_latency_p99_us | 1099 | valid; gateway_c1_samples=7777; direct_c1_samples=244991 | — |
| anthropic → bedrock (upstream over HTTPS) | gateway_c1_p99_us | 1129 | valid; gateway_c1_samples=7777 | — |
| anthropic → bedrock (upstream over HTTPS) | direct_c1_p99_us | 30 | valid; direct_c1_samples=244991 | — |
| anthropic → bedrock (upstream over HTTPS) | cpu_us_per_request | 1001.5123508654008 | valid | — |
| anthropic → bedrock (upstream over HTTPS) | rps_per_cpu_second | 998.489932885906 | valid | — |
| anthropic → bedrock (upstream over HTTPS) | cost_window_conc | 8 | valid | — |
| anthropic → bedrock (upstream over HTTPS) | cost_core_utilisation | 0.9899525457344274 | valid | — |
| anthropic → bedrock (upstream over HTTPS) | cost_window_ok | 23804 | valid | — |
| anthropic → bedrock (upstream over HTTPS) | cost_window_rps | 3966 | valid | — |
| anthropic → bedrock (upstream over HTTPS) | cost_threads | 20 | valid | — |
| anthropic → bedrock (upstream over HTTPS) | cost_nonvol_ctxt_per_request | 0.03297765081498908 | valid | — |
| anthropic → bedrock (upstream over HTTPS) | cost_majflt | 0 | valid | — |
| anthropic → bedrock (upstream over HTTPS) | added_ttft_p50_us | untestable | untestable | — |
| anthropic → bedrock (upstream over HTTPS) | added_ttft_p99_us | untestable | untestable | — |
| anthropic → bedrock (upstream over HTTPS) | added_gap_p50_us | untestable | untestable | — |
| anthropic → bedrock (upstream over HTTPS) | added_gap_p99_us | untestable | untestable | — |
| anthropic → bedrock (upstream over HTTPS) | streams_sustained_fps | untestable | untestable | — |
| anthropic → bedrock (upstream over HTTPS) | streams_sustained | untestable | untestable | — |
| anthropic → bedrock (upstream over HTTPS) | idle_rss_mib | 160.77734375 | valid | — |
| anthropic → bedrock (upstream over HTTPS) | steady_state_rss_mib | 950.16015625 | valid | — |
| anthropic → bedrock (upstream over HTTPS) | recovered_rss_mib | 1053.96875 | valid | — |
| anthropic → bedrock (upstream over HTTPS) | peak_rss_mib | 1191.13671875 | valid | — |
| anthropic → bedrock (upstream over HTTPS) | peak_rss_hwm_mib | 1138.20703125 | valid | — |
| anthropic → bedrock (upstream over HTTPS) | time_to_plateau_s | not_measured | not_measured | — |
| anthropic → bedrock (upstream over HTTPS) | growth_rate_mib_per_min | 99.68931391092528 | valid | — |
| anthropic → bedrock (upstream over HTTPS) | load_s | 300 | valid | — |
| anthropic → bedrock (upstream over HTTPS) | shape | 1 | valid | — |
| anthropic → bedrock (upstream over HTTPS) | idle_shape | not_measured | not_measured | — |
| anthropic → bedrock (upstream over HTTPS) | idle_static | 1 | valid | — |
| anthropic → bedrock (upstream over HTTPS) | idle_growth_rate_mib_per_min | 0.3528148459708091 | valid | — |
| anthropic → bedrock (upstream over HTTPS) | throughput_rps | bound_not_met req/s | 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 |
| anthropic → bedrock (upstream over HTTPS) | throughput_rps | 3426 req/s | concurrency 4; samples 20565 | 5 |
| anthropic → bedrock (upstream over HTTPS) | throughput_rps | 3624 req/s | concurrency 8; samples 21752 | 10 |
| anthropic → bedrock (upstream over HTTPS) | throughput_rps | 3706 req/s | concurrency 32; samples 22241 | 50 |
| anthropic → bedrock (upstream over HTTPS) | throughput_rps | 3706 req/s | concurrency 32; samples 22241 | 100 |
| anthropic → bedrock (upstream over HTTPS) | throughput_rps | 3706 req/s | concurrency 32; samples 22241 | unbounded |
| gemini → bedrock (upstream over HTTPS) | added_latency_p50_us | 732 | valid; gateway_c1_samples=7747; direct_c1_samples=257445 | — |
| gemini → bedrock (upstream over HTTPS) | added_latency_p99_us | 1096 | valid; gateway_c1_samples=7747; direct_c1_samples=257445 | — |
| gemini → bedrock (upstream over HTTPS) | gateway_c1_p99_us | 1125 | valid; gateway_c1_samples=7747 | — |
| gemini → bedrock (upstream over HTTPS) | direct_c1_p99_us | 29 | valid; direct_c1_samples=257445 | — |
| gemini → bedrock (upstream over HTTPS) | cpu_us_per_request | 1022.8790555387995 | valid | — |
| gemini → bedrock (upstream over HTTPS) | rps_per_cpu_second | 977.6326874473464 | valid | — |
| gemini → bedrock (upstream over HTTPS) | cost_window_conc | 8 | valid | — |
| gemini → bedrock (upstream over HTTPS) | cost_core_utilisation | 0.9859755787262404 | valid | — |
| gemini → bedrock (upstream over HTTPS) | cost_window_ok | 23209 | valid | — |
| gemini → bedrock (upstream over HTTPS) | cost_window_rps | 3867 | valid | — |
| gemini → bedrock (upstream over HTTPS) | cost_threads | 13 | valid | — |
| gemini → bedrock (upstream over HTTPS) | cost_nonvol_ctxt_per_request | 0.01930285665043733 | valid | — |
| gemini → bedrock (upstream over HTTPS) | cost_majflt | 0 | valid | — |
| gemini → bedrock (upstream over HTTPS) | added_ttft_p50_us | untestable | untestable | — |
| gemini → bedrock (upstream over HTTPS) | added_ttft_p99_us | untestable | untestable | — |
| gemini → bedrock (upstream over HTTPS) | added_gap_p50_us | untestable | untestable | — |
| gemini → bedrock (upstream over HTTPS) | added_gap_p99_us | untestable | untestable | — |
| gemini → bedrock (upstream over HTTPS) | streams_sustained_fps | untestable | untestable | — |
| gemini → bedrock (upstream over HTTPS) | streams_sustained | untestable | untestable | — |
| gemini → bedrock (upstream over HTTPS) | idle_rss_mib | 159.62890625 | valid | — |
| gemini → bedrock (upstream over HTTPS) | steady_state_rss_mib | 951.703125 | valid | — |
| gemini → bedrock (upstream over HTTPS) | recovered_rss_mib | 1257.5859375 | valid | — |
| gemini → bedrock (upstream over HTTPS) | peak_rss_mib | 1257.6171875 | valid | — |
| gemini → bedrock (upstream over HTTPS) | peak_rss_hwm_mib | 1150.5234375 | valid | — |
| gemini → bedrock (upstream over HTTPS) | time_to_plateau_s | not_measured | not_measured | — |
| gemini → bedrock (upstream over HTTPS) | growth_rate_mib_per_min | 113.76387607449239 | valid | — |
| gemini → bedrock (upstream over HTTPS) | load_s | 300 | valid | — |
| gemini → bedrock (upstream over HTTPS) | shape | 1 | valid | — |
| gemini → bedrock (upstream over HTTPS) | idle_shape | not_measured | not_measured | — |
| gemini → bedrock (upstream over HTTPS) | idle_static | 1 | valid | — |
| gemini → bedrock (upstream over HTTPS) | idle_growth_rate_mib_per_min | 0.38908520290504295 | valid | — |
| gemini → bedrock (upstream over HTTPS) | throughput_rps | bound_not_met req/s | 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 |
| gemini → bedrock (upstream over HTTPS) | throughput_rps | 3839 req/s | concurrency 4; samples 23038 | 5 |
| gemini → bedrock (upstream over HTTPS) | throughput_rps | 3970 req/s | concurrency 8; samples 23829 | 10 |
| gemini → bedrock (upstream over HTTPS) | throughput_rps | 4018 req/s | concurrency 32; samples 24117 | 50 |
| gemini → bedrock (upstream over HTTPS) | throughput_rps | 4018 req/s | concurrency 32; samples 24117 | 100 |
| gemini → bedrock (upstream over HTTPS) | throughput_rps | 4018 req/s | concurrency 32; samples 24117 | unbounded |
| cohere → bedrock (upstream over HTTPS) | added_latency_p50_us | 692 | valid; gateway_c1_samples=8190; direct_c1_samples=262997 | — |
| cohere → bedrock (upstream over HTTPS) | added_latency_p99_us | 1061 | valid; gateway_c1_samples=8190; direct_c1_samples=262997 | — |
| cohere → bedrock (upstream over HTTPS) | gateway_c1_p99_us | 1090 | valid; gateway_c1_samples=8190 | — |
| cohere → bedrock (upstream over HTTPS) | direct_c1_p99_us | 29 | valid; direct_c1_samples=262997 | — |
| cohere → bedrock (upstream over HTTPS) | cpu_us_per_request | 938.6052056427578 | valid | — |
| cohere → bedrock (upstream over HTTPS) | rps_per_cpu_second | 1065.41066892464 | valid | — |
| cohere → bedrock (upstream over HTTPS) | cost_window_conc | 8 | valid | — |
| cohere → bedrock (upstream over HTTPS) | cost_core_utilisation | 0.9810888224468881 | valid | — |
| cohere → bedrock (upstream over HTTPS) | cost_window_ok | 25165 | valid | — |
| cohere → bedrock (upstream over HTTPS) | cost_window_rps | 4193 | valid | — |
| cohere → bedrock (upstream over HTTPS) | cost_threads | 13 | valid | — |
| cohere → bedrock (upstream over HTTPS) | cost_nonvol_ctxt_per_request | 0 | valid | — |
| cohere → bedrock (upstream over HTTPS) | cost_majflt | 0 | valid | — |
| cohere → bedrock (upstream over HTTPS) | added_ttft_p50_us | untestable | untestable | — |
| cohere → bedrock (upstream over HTTPS) | added_ttft_p99_us | untestable | untestable | — |
| cohere → bedrock (upstream over HTTPS) | added_gap_p50_us | untestable | untestable | — |
| cohere → bedrock (upstream over HTTPS) | added_gap_p99_us | untestable | untestable | — |
| cohere → bedrock (upstream over HTTPS) | streams_sustained_fps | untestable | untestable | — |
| cohere → bedrock (upstream over HTTPS) | streams_sustained | untestable | untestable | — |
| cohere → bedrock (upstream over HTTPS) | idle_rss_mib | 162.12890625 | valid | — |
| cohere → bedrock (upstream over HTTPS) | steady_state_rss_mib | 1001.58203125 | valid | — |
| cohere → bedrock (upstream over HTTPS) | recovered_rss_mib | 1040.97265625 | valid | — |
| cohere → bedrock (upstream over HTTPS) | peak_rss_mib | 1135.625 | valid | — |
| cohere → bedrock (upstream over HTTPS) | peak_rss_hwm_mib | 1107.0546875 | valid | — |
| cohere → bedrock (upstream over HTTPS) | time_to_plateau_s | not_measured | not_measured | — |
| cohere → bedrock (upstream over HTTPS) | growth_rate_mib_per_min | 66.49416518170766 | valid | — |
| cohere → bedrock (upstream over HTTPS) | load_s | 300 | valid | — |
| cohere → bedrock (upstream over HTTPS) | shape | 1 | valid | — |
| cohere → bedrock (upstream over HTTPS) | idle_shape | not_measured | not_measured | — |
| cohere → bedrock (upstream over HTTPS) | idle_static | 1 | valid | — |
| cohere → bedrock (upstream over HTTPS) | idle_growth_rate_mib_per_min | 0.7986957676605473 | valid | — |
| cohere → bedrock (upstream over HTTPS) | throughput_rps | bound_not_met req/s | 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 |
| cohere → bedrock (upstream over HTTPS) | throughput_rps | 4119 req/s | concurrency 4; samples 24719 | 5 |
| cohere → bedrock (upstream over HTTPS) | throughput_rps | 4263 req/s | concurrency 8; samples 25581 | 10 |
| cohere → bedrock (upstream over HTTPS) | throughput_rps | 4275 req/s | concurrency 32; samples 25658 | 50 |
| cohere → bedrock (upstream over HTTPS) | throughput_rps | 4275 req/s | concurrency 32; samples 25658 | 100 |
| cohere → bedrock (upstream over HTTPS) | throughput_rps | 4275 req/s | concurrency 32; samples 25658 | unbounded |
| bedrock → bedrock (upstream over HTTPS) | added_latency_p50_us | 694 | valid; gateway_c1_samples=8133; direct_c1_samples=263582 | — |
| bedrock → bedrock (upstream over HTTPS) | added_latency_p99_us | 1052 | valid; gateway_c1_samples=8133; direct_c1_samples=263582 | — |
| bedrock → bedrock (upstream over HTTPS) | gateway_c1_p99_us | 1079 | valid; gateway_c1_samples=8133 | — |
| bedrock → bedrock (upstream over HTTPS) | direct_c1_p99_us | 27 | valid; direct_c1_samples=263582 | — |
| bedrock → bedrock (upstream over HTTPS) | cpu_us_per_request | 956.7628385437596 | valid | — |
| bedrock → bedrock (upstream over HTTPS) | rps_per_cpu_second | 1045.1910961780766 | valid | — |
| bedrock → bedrock (upstream over HTTPS) | cost_window_conc | 8 | valid | — |
| bedrock → bedrock (upstream over HTTPS) | cost_core_utilisation | 0.988885545960124 | valid | — |
| bedrock → bedrock (upstream over HTTPS) | cost_window_ok | 24886 | valid | — |
| bedrock → bedrock (upstream over HTTPS) | cost_window_rps | 4147 | valid | — |
| bedrock → bedrock (upstream over HTTPS) | cost_threads | 14 | valid | — |
| bedrock → bedrock (upstream over HTTPS) | cost_nonvol_ctxt_per_request | 0.03359318492325002 | valid | — |
| bedrock → bedrock (upstream over HTTPS) | cost_majflt | 0 | valid | — |
| bedrock → bedrock (upstream over HTTPS) | added_ttft_p50_us | untestable | untestable | — |
| bedrock → bedrock (upstream over HTTPS) | added_ttft_p99_us | untestable | untestable | — |
| bedrock → bedrock (upstream over HTTPS) | added_gap_p50_us | untestable | untestable | — |
| bedrock → bedrock (upstream over HTTPS) | added_gap_p99_us | untestable | untestable | — |
| bedrock → bedrock (upstream over HTTPS) | streams_sustained_fps | untestable | untestable | — |
| bedrock → bedrock (upstream over HTTPS) | streams_sustained | untestable | untestable | — |
| bedrock → bedrock (upstream over HTTPS) | idle_rss_mib | 160.0234375 | valid | — |
| bedrock → bedrock (upstream over HTTPS) | steady_state_rss_mib | 964.095703125 | valid | — |
| bedrock → bedrock (upstream over HTTPS) | recovered_rss_mib | 1242.09765625 | valid | — |
| bedrock → bedrock (upstream over HTTPS) | peak_rss_mib | 1242.140625 | valid | — |
| bedrock → bedrock (upstream over HTTPS) | peak_rss_hwm_mib | 1119.36328125 | valid | — |
| bedrock → bedrock (upstream over HTTPS) | time_to_plateau_s | not_measured | not_measured | — |
| bedrock → bedrock (upstream over HTTPS) | growth_rate_mib_per_min | 61.39766170077196 | valid | — |
| bedrock → bedrock (upstream over HTTPS) | load_s | 300 | valid | — |
| bedrock → bedrock (upstream over HTTPS) | shape | 1 | valid | — |
| bedrock → bedrock (upstream over HTTPS) | idle_shape | not_measured | not_measured | — |
| bedrock → bedrock (upstream over HTTPS) | idle_static | 1 | valid | — |
| bedrock → bedrock (upstream over HTTPS) | idle_growth_rate_mib_per_min | 0.7829862318956637 | valid | — |
| bedrock → bedrock (upstream over HTTPS) | throughput_rps | bound_not_met req/s | 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 |
| bedrock → bedrock (upstream over HTTPS) | throughput_rps | 3832 req/s | concurrency 4; samples 22999 | 5 |
| bedrock → bedrock (upstream over HTTPS) | throughput_rps | 3972 req/s | concurrency 8; samples 23835 | 10 |
| bedrock → bedrock (upstream over HTTPS) | throughput_rps | 4035 req/s | concurrency 32; samples 24214 | 50 |
| bedrock → bedrock (upstream over HTTPS) | throughput_rps | 4035 req/s | concurrency 32; samples 24214 | 100 |
| bedrock → bedrock (upstream over HTTPS) | throughput_rps | 4035 req/s | concurrency 32; samples 24214 | 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@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| openai-responses → openai | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| anthropic → openai | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| gemini → openai | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| cohere → openai | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| bedrock → openai | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| openai → openai-responses | not_declared | — | [recorded note 1](https://onthebench.ai/#note-1) | [recorded note 1](https://onthebench.ai/#note-1) | [recorded note 2](https://onthebench.ai/#note-2) | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| openai-responses → openai-responses | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| anthropic → openai-responses | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| gemini → openai-responses | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| cohere → openai-responses | not_declared | — | [recorded note 1](https://onthebench.ai/#note-1) | [recorded note 1](https://onthebench.ai/#note-1) | [recorded note 2](https://onthebench.ai/#note-2) | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| bedrock → openai-responses | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| openai → anthropic | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| openai-responses → anthropic | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| anthropic → anthropic | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| gemini → anthropic | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| cohere → anthropic | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| bedrock → anthropic | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| openai → gemini | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| openai-responses → gemini | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| anthropic → gemini | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| gemini → gemini | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| cohere → gemini | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| bedrock → gemini | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| openai → cohere | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| openai-responses → cohere | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| anthropic → cohere | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| gemini → cohere | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| cohere → cohere | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| bedrock → cohere | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| openai → bedrock | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| openai-responses → bedrock | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| anthropic → bedrock | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| gemini → bedrock | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| cohere → bedrock | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |
| bedrock → bedrock | served | — | — | — | Recorded run evidence | — | maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78 |

**Harness commit**: `d240badd966933fa19eec0606b2d1f3945f18ad2`

**Mock SHA-256**: `99d01f709d8d9c1f49055aa265d5eb7475264d6b513911fe317aee74a71190f8`

**Load generator SHA-256**: `not recorded`

**somaxconn**: 65535

**tcp_max_syn_backlog**: 65535

## 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 | {"200":1} | 0 |
| openai → openai | 2 | {"200":2} | 0 |
| openai → openai | 4 | {"200":4} | 0 |
| openai → openai | 8 | {"200":8} | 0 |
| openai → openai | 16 | {"200":16} | 0 |
| openai → openai | 32 | {"200":32} | 0 |
| openai → openai | 64 | {"200":64} | 0 |
| openai → openai | 128 | {"200":128} | 0 |
| openai → openai | 256 | {"200":256} | 0 |
| openai → openai | 512 | {"200":512} | 0 |
| openai → openai | 1024 | {"200":1024} | 0 |
| openai → openai | 768 | {"200":768} | 0 |
| openai → openai | 640 | {"200":640} | 0 |
| openai → openai | 576 | {"200":576} | 0 |
| openai → openai | 608 | {"200":608} | 0 |
| openai → openai | 592 | {"200":592} | 0 |
| openai → openai | 584 | {"200":584} | 0 |
| openai → openai | 580 | {"200":580} | 0 |
| openai → openai | 582 | {"200":582} | 0 |
| openai → openai | 581 | {"200":581} | 0 |
| openai → openai | 580 | {"200":580} | 0 |
| openai → openai | 580 | {"200":580} | 0 |
| anthropic → openai | 1 | {"200":1} | 0 |
| anthropic → openai | 2 | {"200":2} | 0 |
| anthropic → openai | 4 | {"200":4} | 0 |
| anthropic → openai | 8 | {"200":8} | 0 |
| anthropic → openai | 16 | {"200":16} | 0 |
| anthropic → openai | 32 | {"200":32} | 0 |
| anthropic → openai | 64 | {"200":64} | 0 |
| anthropic → openai | 128 | {"200":128} | 0 |
| anthropic → openai | 256 | {"200":256} | 0 |
| anthropic → openai | 512 | {"200":512} | 0 |
| anthropic → openai | 1024 | {"200":1024} | 0 |
| anthropic → openai | 768 | {"200":768} | 0 |
| anthropic → openai | 640 | {"200":640} | 0 |
| anthropic → openai | 704 | {"200":704} | 0 |
| anthropic → openai | 672 | {"200":672} | 0 |
| anthropic → openai | 688 | {"200":688} | 0 |
| anthropic → openai | 680 | {"200":680} | 0 |
| anthropic → openai | 676 | {"200":676} | 0 |
| anthropic → openai | 678 | {"200":678} | 0 |
| anthropic → openai | 679 | {"200":679} | 0 |
| anthropic → openai | 678 | {"200":678} | 0 |
| anthropic → openai | 678 | {"200":678} | 0 |
| openai → anthropic | 1 | {"200":1} | 0 |
| openai → anthropic | 2 | {"200":2} | 0 |
| openai → anthropic | 4 | {"200":4} | 0 |
| openai → anthropic | 8 | {"200":8} | 0 |
| openai → anthropic | 16 | {"200":16} | 0 |
| openai → anthropic | 32 | {"200":32} | 0 |
| openai → anthropic | 64 | {"200":64} | 0 |
| openai → anthropic | 128 | {"200":128} | 0 |
| openai → anthropic | 256 | {"200":256} | 0 |
| openai → anthropic | 512 | {"200":512} | 0 |
| openai → anthropic | 1024 | {"200":1024} | 0 |
| openai → anthropic | 768 | {"200":768} | 0 |
| openai → anthropic | 640 | {"200":640} | 0 |
| openai → anthropic | 576 | {"200":576} | 0 |
| openai → anthropic | 608 | {"200":608} | 0 |
| openai → anthropic | 624 | {"200":624} | 0 |
| openai → anthropic | 632 | {"200":632} | 0 |
| openai → anthropic | 628 | {"200":628} | 0 |
| openai → anthropic | 626 | {"200":626} | 0 |
| openai → anthropic | 625 | {"200":625} | 0 |
| openai → anthropic | 624 | {"200":624} | 0 |
| openai → anthropic | 624 | {"200":624} | 0 |
| anthropic → anthropic | 1 | {"200":1} | 0 |
| anthropic → anthropic | 2 | {"200":2} | 0 |
| anthropic → anthropic | 4 | {"200":4} | 0 |
| anthropic → anthropic | 8 | {"200":8} | 0 |
| anthropic → anthropic | 16 | {"200":16} | 0 |
| anthropic → anthropic | 32 | {"200":32} | 0 |
| anthropic → anthropic | 64 | {"200":64} | 0 |
| anthropic → anthropic | 128 | {"200":128} | 0 |
| anthropic → anthropic | 256 | {"200":256} | 0 |
| anthropic → anthropic | 512 | {"200":512} | 0 |
| anthropic → anthropic | 1024 | {"200":1024} | 0 |
| anthropic → anthropic | 768 | {"200":768} | 0 |
| anthropic → anthropic | 896 | {"200":896} | 0 |
| anthropic → anthropic | 832 | {"200":832} | 0 |
| anthropic → anthropic | 864 | {"200":864} | 0 |
| anthropic → anthropic | 848 | {"200":848} | 0 |
| anthropic → anthropic | 856 | {"200":856} | 0 |
| anthropic → anthropic | 852 | {"200":852} | 0 |
| anthropic → anthropic | 854 | {"200":854} | 0 |
| anthropic → anthropic | 855 | {"200":855} | 0 |
| anthropic → anthropic | 855 | {"200":855} | 0 |
| anthropic → anthropic | 855 | {"200":855} | 0 |
| anthropic → anthropic | 683 | {"200":683} | 0 |
| anthropic → anthropic | 683 | {"200":683} | 0 |
| anthropic → anthropic | 683 | {"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](https://onthebench.ai/gateways/runs/run-bifrost-c253d4668350f1793f587661b425702d5448ca78d04ec9144589514b6064ebd7)  ·  [complete raw record including RSS series](https://onthebench.ai/api/v1/runs/run-bifrost-c253d4668350f1793f587661b425702d5448ca78d04ec9144589514b6064ebd7/raw.json)

## 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](https://onthebench.ai/gateways/compare/bifrost-vs-agentgateway)
- [Bifrost vs AISIX (api7)](https://onthebench.ai/gateways/compare/bifrost-vs-aisix)
- [Bifrost vs APISIX](https://onthebench.ai/gateways/compare/bifrost-vs-apisix)
- [Bifrost vs Busbar](https://onthebench.ai/gateways/compare/bifrost-vs-busbar)
- [Bifrost vs GoModel](https://onthebench.ai/gateways/compare/bifrost-vs-gomodel)
- [Bifrost vs Helicone](https://onthebench.ai/gateways/compare/bifrost-vs-helicone)
- [Bifrost vs Higress](https://onthebench.ai/gateways/compare/bifrost-vs-higress)
- [Bifrost vs Kong](https://onthebench.ai/gateways/compare/bifrost-vs-kong)
- [Bifrost vs LiteLLM · Python](https://onthebench.ai/gateways/compare/bifrost-vs-litellm-python)
- [Bifrost vs LiteLLM · Rust](https://onthebench.ai/gateways/compare/bifrost-vs-litellm-rust)
- [Bifrost vs One-API](https://onthebench.ai/gateways/compare/bifrost-vs-one-api)
- [Bifrost vs Plano](https://onthebench.ai/gateways/compare/bifrost-vs-plano)
- [Bifrost vs Portkey](https://onthebench.ai/gateways/compare/bifrost-vs-portkey)
- [Bifrost vs TensorZero](https://onthebench.ai/gateways/compare/bifrost-vs-tensorzero)

## Historical measurements

- [2026-11-11T03:17:59Z — maximhq/bifrost:v2.5.0@sha256:445bc49544bc64971b63581a93c6ca5716a03ccf27a58ced4c285cbafc133e78](https://onthebench.ai/gateways/runs/run-bifrost-c253d4668350f1793f587661b425702d5448ca78d04ec9144589514b6064ebd7)
- [2026-08-26T19:08:46Z — maximhq/bifrost:v1.6.11](https://onthebench.ai/gateways/runs/run-bifrost-9b2b5f70d61994ac469a86d5949dc62ffadcdb5dfcd808ca2c1a0c0f08368019)
