# Higress 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) .

Other · recorded build: higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422. Measured: 2026-11-10T17:39:42Z.

[JSON](https://onthebench.ai/api/v1/gateways/higress)  ·  [Markdown](https://onthebench.ai/gateways/higress.md)  ·  [CSV](https://onthebench.ai/gateways/data.csv?gateway=higress&compact=1)  ·  [Project source](https://github.com/alibaba/higress)  ·  [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) |
| --- | --- | --- | --- | --- |
| cohere → cohere | added_latency_p50_us | 615 | valid; gateway_c1_samples=9341; direct_c1_samples=267984 | — |
| cohere → cohere | added_latency_p99_us | 747 | valid; gateway_c1_samples=9341; direct_c1_samples=267984 | — |
| cohere → cohere | gateway_c1_p99_us | 774 | valid; gateway_c1_samples=9341 | — |
| cohere → cohere | direct_c1_p99_us | 27 | valid; direct_c1_samples=267984 | — |
| cohere → cohere | cpu_us_per_request | 613.2194813409235 | valid | — |
| cohere → cohere | rps_per_cpu_second | 1630.737493553378 | valid | — |
| cohere → cohere | cost_window_conc | 8 | valid | — |
| cohere → cohere | cost_core_utilisation | 0.8053044884180178 | valid | — |
| cohere → cohere | cost_window_ok | 31620 | valid | — |
| cohere → cohere | cost_window_rps | 5269 | valid | — |
| cohere → cohere | cost_threads | 143 | valid | — |
| cohere → cohere | cost_nonvol_ctxt_per_request | 0.0076217583807716635 | 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 | 1207.94921875 | valid | — |
| cohere → cohere | steady_state_rss_mib | 1217.2890625 | valid | — |
| cohere → cohere | recovered_rss_mib | 1218.421875 | valid | — |
| cohere → cohere | peak_rss_mib | 1226.8046875 | valid | — |
| cohere → cohere | peak_rss_hwm_mib | 1324.41015625 | valid | — |
| cohere → cohere | time_to_plateau_s | 60 | valid | — |
| cohere → cohere | growth_rate_mib_per_min | 0.7482121208635604 | valid | — |
| cohere → cohere | load_s | 300 | valid | — |
| cohere → cohere | shape | not_measured | not_measured | — |
| cohere → cohere | idle_shape | not_measured | not_measured | — |
| cohere → cohere | idle_static | 1 | valid | — |
| cohere → cohere | idle_growth_rate_mib_per_min | 1.0318152244070045 | valid | — |
| cohere → cohere | throughput_rps | 3092 req/s | concurrency 2; samples 18555 | 1 |
| cohere → cohere | throughput_rps | 5921 req/s | concurrency 8; samples 35536 | 5 |
| cohere → cohere | throughput_rps | 6529 req/s | concurrency 32; samples 39189 | 10 |
| cohere → cohere | throughput_rps | 6633 req/s | concurrency 64; samples 39805 | 50 |
| cohere → cohere | throughput_rps | 6633 req/s | concurrency 64; samples 39805 | 100 |
| cohere → cohere | throughput_rps | 6633 req/s | concurrency 64; samples 39805 | unbounded |
| bedrock → cohere | — | — | not_declared: Higress 2.2.5 with bundled AI Proxy 2.0.3. Source 4deae1dd3b4cb2707273748985e55bef28e747d1 and pinned-image probes establish 13 configured exact-wire pairs; two Cohere translations failed response correctness in the preflight. In the 2026-11-09 run the engine withheld performance and streaming numbers for six configured pairs because re-verification found the request did not arrive on the requested upstream wire: OpenAI to Anthropic, Gemini, Cohere and Bedrock arrived untranslated on the mock's OpenAI Chat Completions endpoint, and Anthropic to Gemini and Anthropic to Cohere also arrived on the OpenAI Chat Completions endpoint. Responses cross-conversion is not implemented; Messages-to-Bedrock uses Anthropic Mantle rather than Converse. Pairs outside the configured 13 are not configured: unsupported translations pass through the original wire or return unsupported API errors. Evidence: results/diagnostics/refresh-configuration/higress-adapter-preflight.json | — |
| gemini → cohere | — | — | not_declared: Higress 2.2.5 with bundled AI Proxy 2.0.3. Source 4deae1dd3b4cb2707273748985e55bef28e747d1 and pinned-image probes establish 13 configured exact-wire pairs; two Cohere translations failed response correctness in the preflight. In the 2026-11-09 run the engine withheld performance and streaming numbers for six configured pairs because re-verification found the request did not arrive on the requested upstream wire: OpenAI to Anthropic, Gemini, Cohere and Bedrock arrived untranslated on the mock's OpenAI Chat Completions endpoint, and Anthropic to Gemini and Anthropic to Cohere also arrived on the OpenAI Chat Completions endpoint. Responses cross-conversion is not implemented; Messages-to-Bedrock uses Anthropic Mantle rather than Converse. Pairs outside the configured 13 are not configured: unsupported translations pass through the original wire or return unsupported API errors. Evidence: results/diagnostics/refresh-configuration/higress-adapter-preflight.json | — |
| openai → cohere | — | — | not_served: probed and answered HTTP 200, but the body is not a usable openai response (see body_snippet) | — |
| anthropic → cohere | — | — | not_served: probed and answered HTTP 200, but the body is not a usable anthropic response (see body_snippet) | — |
| openai-responses → cohere | — | — | not_declared: Higress 2.2.5 with bundled AI Proxy 2.0.3. Source 4deae1dd3b4cb2707273748985e55bef28e747d1 and pinned-image probes establish 13 configured exact-wire pairs; two Cohere translations failed response correctness in the preflight. In the 2026-11-09 run the engine withheld performance and streaming numbers for six configured pairs because re-verification found the request did not arrive on the requested upstream wire: OpenAI to Anthropic, Gemini, Cohere and Bedrock arrived untranslated on the mock's OpenAI Chat Completions endpoint, and Anthropic to Gemini and Anthropic to Cohere also arrived on the OpenAI Chat Completions endpoint. Responses cross-conversion is not implemented; Messages-to-Bedrock uses Anthropic Mantle rather than Converse. Pairs outside the configured 13 are not configured: unsupported translations pass through the original wire or return unsupported API errors. Evidence: results/diagnostics/refresh-configuration/higress-adapter-preflight.json | — |
| cohere → anthropic | — | — | not_declared: Higress 2.2.5 with bundled AI Proxy 2.0.3. Source 4deae1dd3b4cb2707273748985e55bef28e747d1 and pinned-image probes establish 13 configured exact-wire pairs; two Cohere translations failed response correctness in the preflight. In the 2026-11-09 run the engine withheld performance and streaming numbers for six configured pairs because re-verification found the request did not arrive on the requested upstream wire: OpenAI to Anthropic, Gemini, Cohere and Bedrock arrived untranslated on the mock's OpenAI Chat Completions endpoint, and Anthropic to Gemini and Anthropic to Cohere also arrived on the OpenAI Chat Completions endpoint. Responses cross-conversion is not implemented; Messages-to-Bedrock uses Anthropic Mantle rather than Converse. Pairs outside the configured 13 are not configured: unsupported translations pass through the original wire or return unsupported API errors. Evidence: results/diagnostics/refresh-configuration/higress-adapter-preflight.json | — |
| gemini → anthropic | — | — | not_declared: Higress 2.2.5 with bundled AI Proxy 2.0.3. Source 4deae1dd3b4cb2707273748985e55bef28e747d1 and pinned-image probes establish 13 configured exact-wire pairs; two Cohere translations failed response correctness in the preflight. In the 2026-11-09 run the engine withheld performance and streaming numbers for six configured pairs because re-verification found the request did not arrive on the requested upstream wire: OpenAI to Anthropic, Gemini, Cohere and Bedrock arrived untranslated on the mock's OpenAI Chat Completions endpoint, and Anthropic to Gemini and Anthropic to Cohere also arrived on the OpenAI Chat Completions endpoint. Responses cross-conversion is not implemented; Messages-to-Bedrock uses Anthropic Mantle rather than Converse. Pairs outside the configured 13 are not configured: unsupported translations pass through the original wire or return unsupported API errors. Evidence: results/diagnostics/refresh-configuration/higress-adapter-preflight.json | — |
| openai → anthropic | added_latency_p50_us | 859 | valid; gateway_c1_samples=6739; direct_c1_samples=262868 | — |
| openai → anthropic | added_latency_p99_us | 1112 | valid; gateway_c1_samples=6739; direct_c1_samples=262868 | — |
| openai → anthropic | gateway_c1_p99_us | 1140 | valid; gateway_c1_samples=6739 | — |
| openai → anthropic | direct_c1_p99_us | 28 | valid; direct_c1_samples=262868 | — |
| openai → anthropic | cpu_us_per_request | 882.4409213281484 | valid | — |
| openai → anthropic | rps_per_cpu_second | 1133.2203389830509 | valid | — |
| openai → anthropic | cost_window_conc | 8 | valid | — |
| openai → anthropic | cost_core_utilisation | 0.980004881341195 | valid | — |
| openai → anthropic | cost_window_ok | 26744 | valid | — |
| openai → anthropic | cost_window_rps | 4456 | valid | — |
| openai → anthropic | cost_threads | 145 | valid | — |
| openai → anthropic | cost_nonvol_ctxt_per_request | 0.001121746933891714 | valid | — |
| openai → anthropic | cost_majflt | 0 | valid | — |
| openai → anthropic | added_ttft_p50_us | 1095 | valid; ttft_gw_samples=100; ttft_direct_samples=100 | — |
| openai → anthropic | added_ttft_p99_us | 1556 | valid; ttft_gw_samples=100; ttft_direct_samples=100 | — |
| openai → anthropic | added_gap_p50_us | 8 | valid; gap_gw_samples=63; gap_direct_samples=63 | — |
| openai → anthropic | added_gap_p99_us | below_resolution | below_resolution; gap_gw_samples=63; gap_direct_samples=63 | — |
| openai → anthropic | streams_sustained_fps | 23656.559777267295 | valid | — |
| openai → anthropic | streams_sustained | 5767 | valid | — |
| openai → anthropic | idle_rss_mib | 1148.421875 | valid | — |
| openai → anthropic | steady_state_rss_mib | 1158.875 | valid | — |
| openai → anthropic | recovered_rss_mib | 1159.7734375 | valid | — |
| openai → anthropic | peak_rss_mib | 1167.30859375 | valid | — |
| openai → anthropic | peak_rss_hwm_mib | 1264.890625 | valid | — |
| openai → anthropic | time_to_plateau_s | 60 | valid | — |
| openai → anthropic | growth_rate_mib_per_min | 0.7021128067897227 | valid | — |
| openai → anthropic | load_s | 300 | valid | — |
| openai → anthropic | shape | not_measured | not_measured | — |
| openai → anthropic | idle_shape | not_measured | not_measured | — |
| openai → anthropic | idle_static | 1 | valid | — |
| openai → anthropic | idle_growth_rate_mib_per_min | 0.8890053062912313 | 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 | 4788 req/s | concurrency 16; samples 28736 | 5 |
| openai → anthropic | throughput_rps | 4909 req/s | concurrency 32; samples 29464 | 10 |
| openai → anthropic | throughput_rps | 4944 req/s | concurrency 64; samples 29673 | 50 |
| openai → anthropic | throughput_rps | 4944 req/s | concurrency 64; samples 29673 | 100 |
| openai → anthropic | throughput_rps | 4944 req/s | concurrency 64; samples 29673 | unbounded |
| bedrock → anthropic | — | — | not_declared: Higress 2.2.5 with bundled AI Proxy 2.0.3. Source 4deae1dd3b4cb2707273748985e55bef28e747d1 and pinned-image probes establish 13 configured exact-wire pairs; two Cohere translations failed response correctness in the preflight. In the 2026-11-09 run the engine withheld performance and streaming numbers for six configured pairs because re-verification found the request did not arrive on the requested upstream wire: OpenAI to Anthropic, Gemini, Cohere and Bedrock arrived untranslated on the mock's OpenAI Chat Completions endpoint, and Anthropic to Gemini and Anthropic to Cohere also arrived on the OpenAI Chat Completions endpoint. Responses cross-conversion is not implemented; Messages-to-Bedrock uses Anthropic Mantle rather than Converse. Pairs outside the configured 13 are not configured: unsupported translations pass through the original wire or return unsupported API errors. Evidence: results/diagnostics/refresh-configuration/higress-adapter-preflight.json | — |
| openai-responses → anthropic | — | — | not_declared: Higress 2.2.5 with bundled AI Proxy 2.0.3. Source 4deae1dd3b4cb2707273748985e55bef28e747d1 and pinned-image probes establish 13 configured exact-wire pairs; two Cohere translations failed response correctness in the preflight. In the 2026-11-09 run the engine withheld performance and streaming numbers for six configured pairs because re-verification found the request did not arrive on the requested upstream wire: OpenAI to Anthropic, Gemini, Cohere and Bedrock arrived untranslated on the mock's OpenAI Chat Completions endpoint, and Anthropic to Gemini and Anthropic to Cohere also arrived on the OpenAI Chat Completions endpoint. Responses cross-conversion is not implemented; Messages-to-Bedrock uses Anthropic Mantle rather than Converse. Pairs outside the configured 13 are not configured: unsupported translations pass through the original wire or return unsupported API errors. Evidence: results/diagnostics/refresh-configuration/higress-adapter-preflight.json | — |
| anthropic → anthropic | added_latency_p50_us | 680 | valid; gateway_c1_samples=8481; direct_c1_samples=262245 | — |
| anthropic → anthropic | added_latency_p99_us | 822 | valid; gateway_c1_samples=8481; direct_c1_samples=262245 | — |
| anthropic → anthropic | gateway_c1_p99_us | 850 | valid; gateway_c1_samples=8481 | — |
| anthropic → anthropic | direct_c1_p99_us | 28 | valid; direct_c1_samples=262245 | — |
| anthropic → anthropic | cpu_us_per_request | 674.8307904095445 | valid | — |
| anthropic → anthropic | rps_per_cpu_second | 1481.852953676158 | valid | — |
| anthropic → anthropic | cost_window_conc | 8 | valid | — |
| anthropic → anthropic | cost_core_utilisation | 0.9771911881943663 | valid | — |
| anthropic → anthropic | cost_window_ok | 34868 | valid | — |
| anthropic → anthropic | cost_window_rps | 5810 | valid | — |
| anthropic → anthropic | cost_threads | 144 | valid | — |
| anthropic → anthropic | cost_nonvol_ctxt_per_request | 0.0020075714121830907 | valid | — |
| anthropic → anthropic | cost_majflt | 0 | valid | — |
| anthropic → anthropic | added_ttft_p50_us | 822 | valid; ttft_gw_samples=100; ttft_direct_samples=100 | — |
| anthropic → anthropic | added_ttft_p99_us | 1012 | valid; ttft_gw_samples=100; ttft_direct_samples=100 | — |
| anthropic → anthropic | added_gap_p50_us | 0 | valid; gap_gw_samples=63; gap_direct_samples=63 | — |
| anthropic → anthropic | added_gap_p99_us | 35 | valid; gap_gw_samples=63; gap_direct_samples=63 | — |
| anthropic → anthropic | streams_sustained_fps | 52509.65359935977 | valid | — |
| anthropic → anthropic | streams_sustained | 16391 | valid | — |
| anthropic → anthropic | idle_rss_mib | 1153.921875 | valid | — |
| anthropic → anthropic | steady_state_rss_mib | 1161.83984375 | valid | — |
| anthropic → anthropic | recovered_rss_mib | 1162.21484375 | valid | — |
| anthropic → anthropic | peak_rss_mib | 1171.8125 | valid | — |
| anthropic → anthropic | peak_rss_hwm_mib | 1272.52734375 | valid | — |
| anthropic → anthropic | time_to_plateau_s | 60 | valid | — |
| anthropic → anthropic | growth_rate_mib_per_min | 0.275204255009843 | valid | — |
| anthropic → anthropic | load_s | 300 | valid | — |
| anthropic → anthropic | shape | not_measured | not_measured | — |
| anthropic → anthropic | idle_shape | not_measured | not_measured | — |
| anthropic → anthropic | idle_static | 1 | valid | — |
| anthropic → anthropic | idle_growth_rate_mib_per_min | 0.5817656189913647 | valid | — |
| anthropic → anthropic | throughput_rps | 1352 req/s | concurrency 1; samples 8115 | 1 |
| anthropic → anthropic | throughput_rps | 5967 req/s | concurrency 16; samples 35809 | 5 |
| anthropic → anthropic | throughput_rps | 6123 req/s | concurrency 32; samples 36743 | 10 |
| anthropic → anthropic | throughput_rps | 6224 req/s | concurrency 128; samples 37352 | 50 |
| anthropic → anthropic | throughput_rps | 6224 req/s | concurrency 128; samples 37352 | 100 |
| anthropic → anthropic | throughput_rps | 6224 req/s | concurrency 128; samples 37352 | unbounded |
| openai → openai-responses | — | — | not_declared: Higress 2.2.5 with bundled AI Proxy 2.0.3. Source 4deae1dd3b4cb2707273748985e55bef28e747d1 and pinned-image probes establish 13 configured exact-wire pairs; two Cohere translations failed response correctness in the preflight. In the 2026-11-09 run the engine withheld performance and streaming numbers for six configured pairs because re-verification found the request did not arrive on the requested upstream wire: OpenAI to Anthropic, Gemini, Cohere and Bedrock arrived untranslated on the mock's OpenAI Chat Completions endpoint, and Anthropic to Gemini and Anthropic to Cohere also arrived on the OpenAI Chat Completions endpoint. Responses cross-conversion is not implemented; Messages-to-Bedrock uses Anthropic Mantle rather than Converse. Pairs outside the configured 13 are not configured: unsupported translations pass through the original wire or return unsupported API errors. Evidence: results/diagnostics/refresh-configuration/higress-adapter-preflight.json | — |
| gemini → openai-responses | — | — | not_declared: Higress 2.2.5 with bundled AI Proxy 2.0.3. Source 4deae1dd3b4cb2707273748985e55bef28e747d1 and pinned-image probes establish 13 configured exact-wire pairs; two Cohere translations failed response correctness in the preflight. In the 2026-11-09 run the engine withheld performance and streaming numbers for six configured pairs because re-verification found the request did not arrive on the requested upstream wire: OpenAI to Anthropic, Gemini, Cohere and Bedrock arrived untranslated on the mock's OpenAI Chat Completions endpoint, and Anthropic to Gemini and Anthropic to Cohere also arrived on the OpenAI Chat Completions endpoint. Responses cross-conversion is not implemented; Messages-to-Bedrock uses Anthropic Mantle rather than Converse. Pairs outside the configured 13 are not configured: unsupported translations pass through the original wire or return unsupported API errors. Evidence: results/diagnostics/refresh-configuration/higress-adapter-preflight.json | — |
| cohere → openai-responses | — | — | not_declared: Higress 2.2.5 with bundled AI Proxy 2.0.3. Source 4deae1dd3b4cb2707273748985e55bef28e747d1 and pinned-image probes establish 13 configured exact-wire pairs; two Cohere translations failed response correctness in the preflight. In the 2026-11-09 run the engine withheld performance and streaming numbers for six configured pairs because re-verification found the request did not arrive on the requested upstream wire: OpenAI to Anthropic, Gemini, Cohere and Bedrock arrived untranslated on the mock's OpenAI Chat Completions endpoint, and Anthropic to Gemini and Anthropic to Cohere also arrived on the OpenAI Chat Completions endpoint. Responses cross-conversion is not implemented; Messages-to-Bedrock uses Anthropic Mantle rather than Converse. Pairs outside the configured 13 are not configured: unsupported translations pass through the original wire or return unsupported API errors. Evidence: results/diagnostics/refresh-configuration/higress-adapter-preflight.json | — |
| bedrock → openai-responses | — | — | not_declared: Higress 2.2.5 with bundled AI Proxy 2.0.3. Source 4deae1dd3b4cb2707273748985e55bef28e747d1 and pinned-image probes establish 13 configured exact-wire pairs; two Cohere translations failed response correctness in the preflight. In the 2026-11-09 run the engine withheld performance and streaming numbers for six configured pairs because re-verification found the request did not arrive on the requested upstream wire: OpenAI to Anthropic, Gemini, Cohere and Bedrock arrived untranslated on the mock's OpenAI Chat Completions endpoint, and Anthropic to Gemini and Anthropic to Cohere also arrived on the OpenAI Chat Completions endpoint. Responses cross-conversion is not implemented; Messages-to-Bedrock uses Anthropic Mantle rather than Converse. Pairs outside the configured 13 are not configured: unsupported translations pass through the original wire or return unsupported API errors. Evidence: results/diagnostics/refresh-configuration/higress-adapter-preflight.json | — |
| openai-responses → openai-responses | added_latency_p50_us | 642 | valid; gateway_c1_samples=8971; direct_c1_samples=260207 | — |
| openai-responses → openai-responses | added_latency_p99_us | 747 | valid; gateway_c1_samples=8971; direct_c1_samples=260207 | — |
| openai-responses → openai-responses | gateway_c1_p99_us | 775 | valid; gateway_c1_samples=8971 | — |
| openai-responses → openai-responses | direct_c1_p99_us | 28 | valid; direct_c1_samples=260207 | — |
| openai-responses → openai-responses | cpu_us_per_request | 633.7934164508637 | valid | — |
| openai-responses → openai-responses | rps_per_cpu_second | 1577.8011794439765 | valid | — |
| openai-responses → openai-responses | cost_window_conc | 8 | valid | — |
| openai-responses → openai-responses | cost_core_utilisation | 0.9859729286272246 | valid | — |
| openai-responses → openai-responses | cost_window_ok | 37457 | valid | — |
| openai-responses → openai-responses | cost_window_rps | 6242 | valid | — |
| openai-responses → openai-responses | cost_threads | 141 | valid | — |
| openai-responses → openai-responses | cost_nonvol_ctxt_per_request | 0.0010144966227941374 | 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 | 1147.16796875 | valid | — |
| openai-responses → openai-responses | steady_state_rss_mib | 1156.55078125 | valid | — |
| openai-responses → openai-responses | recovered_rss_mib | 1157.92578125 | valid | — |
| openai-responses → openai-responses | peak_rss_mib | 1165.46875 | valid | — |
| openai-responses → openai-responses | peak_rss_hwm_mib | 1186.5546875 | valid | — |
| openai-responses → openai-responses | time_to_plateau_s | 66 | valid | — |
| openai-responses → openai-responses | growth_rate_mib_per_min | 0.5987504753473016 | valid | — |
| openai-responses → openai-responses | load_s | 300 | valid | — |
| openai-responses → openai-responses | shape | not_measured | not_measured | — |
| 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 | 0.6971452873324683 | valid | — |
| openai-responses → openai-responses | throughput_rps | 2925 req/s | concurrency 2; samples 17553 | 1 |
| openai-responses → openai-responses | throughput_rps | 6438 req/s | concurrency 16; samples 38636 | 5 |
| openai-responses → openai-responses | throughput_rps | 6717 req/s | concurrency 32; samples 40311 | 10 |
| openai-responses → openai-responses | throughput_rps | 6791 req/s | concurrency 128; samples 40754 | 50 |
| openai-responses → openai-responses | throughput_rps | 6791 req/s | concurrency 128; samples 40754 | 100 |
| openai-responses → openai-responses | throughput_rps | 6791 req/s | concurrency 128; samples 40754 | unbounded |
| anthropic → openai-responses | — | — | not_declared: Higress 2.2.5 with bundled AI Proxy 2.0.3. Source 4deae1dd3b4cb2707273748985e55bef28e747d1 and pinned-image probes establish 13 configured exact-wire pairs; two Cohere translations failed response correctness in the preflight. In the 2026-11-09 run the engine withheld performance and streaming numbers for six configured pairs because re-verification found the request did not arrive on the requested upstream wire: OpenAI to Anthropic, Gemini, Cohere and Bedrock arrived untranslated on the mock's OpenAI Chat Completions endpoint, and Anthropic to Gemini and Anthropic to Cohere also arrived on the OpenAI Chat Completions endpoint. Responses cross-conversion is not implemented; Messages-to-Bedrock uses Anthropic Mantle rather than Converse. Pairs outside the configured 13 are not configured: unsupported translations pass through the original wire or return unsupported API errors. Evidence: results/diagnostics/refresh-configuration/higress-adapter-preflight.json | — |
| openai → gemini | added_latency_p50_us | 923 | valid; gateway_c1_samples=6239; direct_c1_samples=265561 | — |
| openai → gemini | added_latency_p99_us | 1226 | valid; gateway_c1_samples=6239; direct_c1_samples=265561 | — |
| openai → gemini | gateway_c1_p99_us | 1254 | valid; gateway_c1_samples=6239 | — |
| openai → gemini | direct_c1_p99_us | 28 | valid; direct_c1_samples=265561 | — |
| openai → gemini | cpu_us_per_request | 951.6544794570992 | valid | — |
| openai → gemini | rps_per_cpu_second | 1050.8015478164732 | valid | — |
| openai → gemini | cost_window_conc | 8 | valid | — |
| openai → gemini | cost_core_utilisation | 0.7511344229974055 | valid | — |
| openai → gemini | cost_window_ok | 19009 | valid | — |
| openai → gemini | cost_window_rps | 3167 | valid | — |
| openai → gemini | cost_threads | 141 | valid | — |
| openai → gemini | cost_nonvol_ctxt_per_request | 0.011047398600662844 | 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 | 1150.5859375 | valid | — |
| openai → gemini | steady_state_rss_mib | 1160.25390625 | valid | — |
| openai → gemini | recovered_rss_mib | 1160.609375 | valid | — |
| openai → gemini | peak_rss_mib | 1170.90234375 | valid | — |
| openai → gemini | peak_rss_hwm_mib | 1268.890625 | valid | — |
| openai → gemini | time_to_plateau_s | 60 | valid | — |
| openai → gemini | growth_rate_mib_per_min | 0.25904735694337694 | valid | — |
| openai → gemini | load_s | 300 | valid | — |
| openai → gemini | shape | not_measured | not_measured | — |
| openai → gemini | idle_shape | not_measured | not_measured | — |
| openai → gemini | idle_static | 1 | valid | — |
| openai → gemini | idle_growth_rate_mib_per_min | 0.6054154815784856 | 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 | 4171 req/s | concurrency 8; samples 25035 | 5 |
| openai → gemini | throughput_rps | 4327 req/s | concurrency 16; samples 25969 | 10 |
| openai → gemini | throughput_rps | 4449 req/s | concurrency 64; samples 26699 | 50 |
| openai → gemini | throughput_rps | 4449 req/s | concurrency 64; samples 26699 | 100 |
| openai → gemini | throughput_rps | 4449 req/s | concurrency 64; samples 26699 | unbounded |
| gemini → gemini | added_latency_p50_us | 750 | valid; gateway_c1_samples=7714; direct_c1_samples=265544 | — |
| gemini → gemini | added_latency_p99_us | 932 | valid; gateway_c1_samples=7714; direct_c1_samples=265544 | — |
| gemini → gemini | gateway_c1_p99_us | 959 | valid; gateway_c1_samples=7714 | — |
| gemini → gemini | direct_c1_p99_us | 27 | valid; direct_c1_samples=265544 | — |
| gemini → gemini | cpu_us_per_request | 737.715063745513 | valid | — |
| gemini → gemini | rps_per_cpu_second | 1355.5369127516778 | valid | — |
| gemini → gemini | cost_window_conc | 8 | valid | — |
| gemini → gemini | cost_core_utilisation | 0.7426132259918599 | valid | — |
| gemini → gemini | cost_window_ok | 24237 | valid | — |
| gemini → gemini | cost_window_rps | 4038 | valid | — |
| gemini → gemini | cost_threads | 140 | valid | — |
| gemini → gemini | cost_nonvol_ctxt_per_request | 0.007014069398027809 | 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 | 1156.921875 | valid | — |
| gemini → gemini | steady_state_rss_mib | 1165.796875 | valid | — |
| gemini → gemini | recovered_rss_mib | 1166.64453125 | valid | — |
| gemini → gemini | peak_rss_mib | 1175.75390625 | valid | — |
| gemini → gemini | peak_rss_hwm_mib | 1269.25 | valid | — |
| gemini → gemini | time_to_plateau_s | 66 | valid | — |
| gemini → gemini | growth_rate_mib_per_min | 1.08419810667365 | valid | — |
| gemini → gemini | load_s | 300 | valid | — |
| gemini → gemini | shape | not_measured | not_measured | — |
| gemini → gemini | idle_shape | not_measured | not_measured | — |
| gemini → gemini | idle_static | 1 | valid | — |
| gemini → gemini | idle_growth_rate_mib_per_min | 0.8945988670581346 | 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 | 5241 req/s | concurrency 16; samples 31452 | 5 |
| gemini → gemini | throughput_rps | 5428 req/s | concurrency 32; samples 32577 | 10 |
| gemini → gemini | throughput_rps | 5506 req/s | concurrency 64; samples 33045 | 50 |
| gemini → gemini | throughput_rps | 5506 req/s | concurrency 64; samples 33045 | 100 |
| gemini → gemini | throughput_rps | 5506 req/s | concurrency 64; samples 33045 | unbounded |
| bedrock → gemini | — | — | not_declared: Higress 2.2.5 with bundled AI Proxy 2.0.3. Source 4deae1dd3b4cb2707273748985e55bef28e747d1 and pinned-image probes establish 13 configured exact-wire pairs; two Cohere translations failed response correctness in the preflight. In the 2026-11-09 run the engine withheld performance and streaming numbers for six configured pairs because re-verification found the request did not arrive on the requested upstream wire: OpenAI to Anthropic, Gemini, Cohere and Bedrock arrived untranslated on the mock's OpenAI Chat Completions endpoint, and Anthropic to Gemini and Anthropic to Cohere also arrived on the OpenAI Chat Completions endpoint. Responses cross-conversion is not implemented; Messages-to-Bedrock uses Anthropic Mantle rather than Converse. Pairs outside the configured 13 are not configured: unsupported translations pass through the original wire or return unsupported API errors. Evidence: results/diagnostics/refresh-configuration/higress-adapter-preflight.json | — |
| cohere → gemini | — | — | not_declared: Higress 2.2.5 with bundled AI Proxy 2.0.3. Source 4deae1dd3b4cb2707273748985e55bef28e747d1 and pinned-image probes establish 13 configured exact-wire pairs; two Cohere translations failed response correctness in the preflight. In the 2026-11-09 run the engine withheld performance and streaming numbers for six configured pairs because re-verification found the request did not arrive on the requested upstream wire: OpenAI to Anthropic, Gemini, Cohere and Bedrock arrived untranslated on the mock's OpenAI Chat Completions endpoint, and Anthropic to Gemini and Anthropic to Cohere also arrived on the OpenAI Chat Completions endpoint. Responses cross-conversion is not implemented; Messages-to-Bedrock uses Anthropic Mantle rather than Converse. Pairs outside the configured 13 are not configured: unsupported translations pass through the original wire or return unsupported API errors. Evidence: results/diagnostics/refresh-configuration/higress-adapter-preflight.json | — |
| openai-responses → gemini | — | — | not_declared: Higress 2.2.5 with bundled AI Proxy 2.0.3. Source 4deae1dd3b4cb2707273748985e55bef28e747d1 and pinned-image probes establish 13 configured exact-wire pairs; two Cohere translations failed response correctness in the preflight. In the 2026-11-09 run the engine withheld performance and streaming numbers for six configured pairs because re-verification found the request did not arrive on the requested upstream wire: OpenAI to Anthropic, Gemini, Cohere and Bedrock arrived untranslated on the mock's OpenAI Chat Completions endpoint, and Anthropic to Gemini and Anthropic to Cohere also arrived on the OpenAI Chat Completions endpoint. Responses cross-conversion is not implemented; Messages-to-Bedrock uses Anthropic Mantle rather than Converse. Pairs outside the configured 13 are not configured: unsupported translations pass through the original wire or return unsupported API errors. Evidence: results/diagnostics/refresh-configuration/higress-adapter-preflight.json | — |
| anthropic → gemini | added_latency_p50_us | 1045 | valid; gateway_c1_samples=5488; direct_c1_samples=262925 | — |
| anthropic → gemini | added_latency_p99_us | 1360 | valid; gateway_c1_samples=5488; direct_c1_samples=262925 | — |
| anthropic → gemini | gateway_c1_p99_us | 1388 | valid; gateway_c1_samples=5488 | — |
| anthropic → gemini | direct_c1_p99_us | 28 | valid; direct_c1_samples=262925 | — |
| anthropic → gemini | cpu_us_per_request | 1055.3296800424253 | valid | — |
| anthropic → gemini | rps_per_cpu_second | 947.571189279732 | valid | — |
| anthropic → gemini | cost_window_conc | 8 | valid | — |
| anthropic → gemini | cost_core_utilisation | 0.7438065009330738 | valid | — |
| anthropic → gemini | cost_window_ok | 16971 | valid | — |
| anthropic → gemini | cost_window_rps | 2828 | valid | — |
| anthropic → gemini | cost_threads | 142 | valid | — |
| anthropic → gemini | cost_nonvol_ctxt_per_request | 0.01019386011431265 | 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 | 1151.53125 | valid | — |
| anthropic → gemini | steady_state_rss_mib | 1164.61328125 | valid | — |
| anthropic → gemini | recovered_rss_mib | 1165.828125 | valid | — |
| anthropic → gemini | peak_rss_mib | 1173.3046875 | valid | — |
| anthropic → gemini | peak_rss_hwm_mib | 1280.23828125 | valid | — |
| anthropic → gemini | time_to_plateau_s | 60 | valid | — |
| anthropic → gemini | growth_rate_mib_per_min | 1.1668804149685401 | valid | — |
| anthropic → gemini | load_s | 300 | valid | — |
| anthropic → gemini | shape | not_measured | not_measured | — |
| anthropic → gemini | idle_shape | not_measured | not_measured | — |
| anthropic → gemini | idle_static | 1 | valid | — |
| anthropic → gemini | idle_growth_rate_mib_per_min | 1.20687057093733 | 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 | 3821 req/s | concurrency 8; samples 22929 | 5 |
| anthropic → gemini | throughput_rps | 3862 req/s | concurrency 16; samples 23185 | 10 |
| anthropic → gemini | throughput_rps | 3961 req/s | concurrency 64; samples 23771 | 50 |
| anthropic → gemini | throughput_rps | 3961 req/s | concurrency 64; samples 23771 | 100 |
| anthropic → gemini | throughput_rps | 3961 req/s | concurrency 64; samples 23771 | unbounded |
| openai-responses → bedrock | — | — | not_declared: Higress 2.2.5 with bundled AI Proxy 2.0.3. Source 4deae1dd3b4cb2707273748985e55bef28e747d1 and pinned-image probes establish 13 configured exact-wire pairs; two Cohere translations failed response correctness in the preflight. In the 2026-11-09 run the engine withheld performance and streaming numbers for six configured pairs because re-verification found the request did not arrive on the requested upstream wire: OpenAI to Anthropic, Gemini, Cohere and Bedrock arrived untranslated on the mock's OpenAI Chat Completions endpoint, and Anthropic to Gemini and Anthropic to Cohere also arrived on the OpenAI Chat Completions endpoint. Responses cross-conversion is not implemented; Messages-to-Bedrock uses Anthropic Mantle rather than Converse. Pairs outside the configured 13 are not configured: unsupported translations pass through the original wire or return unsupported API errors. Evidence: results/diagnostics/refresh-configuration/higress-adapter-preflight.json | — |
| gemini → bedrock | — | — | not_declared: Higress 2.2.5 with bundled AI Proxy 2.0.3. Source 4deae1dd3b4cb2707273748985e55bef28e747d1 and pinned-image probes establish 13 configured exact-wire pairs; two Cohere translations failed response correctness in the preflight. In the 2026-11-09 run the engine withheld performance and streaming numbers for six configured pairs because re-verification found the request did not arrive on the requested upstream wire: OpenAI to Anthropic, Gemini, Cohere and Bedrock arrived untranslated on the mock's OpenAI Chat Completions endpoint, and Anthropic to Gemini and Anthropic to Cohere also arrived on the OpenAI Chat Completions endpoint. Responses cross-conversion is not implemented; Messages-to-Bedrock uses Anthropic Mantle rather than Converse. Pairs outside the configured 13 are not configured: unsupported translations pass through the original wire or return unsupported API errors. Evidence: results/diagnostics/refresh-configuration/higress-adapter-preflight.json | — |
| cohere → bedrock | — | — | not_declared: Higress 2.2.5 with bundled AI Proxy 2.0.3. Source 4deae1dd3b4cb2707273748985e55bef28e747d1 and pinned-image probes establish 13 configured exact-wire pairs; two Cohere translations failed response correctness in the preflight. In the 2026-11-09 run the engine withheld performance and streaming numbers for six configured pairs because re-verification found the request did not arrive on the requested upstream wire: OpenAI to Anthropic, Gemini, Cohere and Bedrock arrived untranslated on the mock's OpenAI Chat Completions endpoint, and Anthropic to Gemini and Anthropic to Cohere also arrived on the OpenAI Chat Completions endpoint. Responses cross-conversion is not implemented; Messages-to-Bedrock uses Anthropic Mantle rather than Converse. Pairs outside the configured 13 are not configured: unsupported translations pass through the original wire or return unsupported API errors. Evidence: results/diagnostics/refresh-configuration/higress-adapter-preflight.json | — |
| bedrock → bedrock | added_latency_p50_us | 727 | valid; gateway_c1_samples=7954; direct_c1_samples=263620 | — |
| bedrock → bedrock | added_latency_p99_us | 861 | valid; gateway_c1_samples=7954; direct_c1_samples=263620 | — |
| bedrock → bedrock | gateway_c1_p99_us | 889 | valid; gateway_c1_samples=7954 | — |
| bedrock → bedrock | direct_c1_p99_us | 28 | valid; direct_c1_samples=263620 | — |
| bedrock → bedrock | cpu_us_per_request | 778.1517645882627 | valid | — |
| bedrock → bedrock | rps_per_cpu_second | 1285.0963597430407 | valid | — |
| bedrock → bedrock | cost_window_conc | 8 | valid | — |
| bedrock → bedrock | cost_core_utilisation | 0.9697792765122447 | valid | — |
| bedrock → bedrock | cost_window_ok | 30007 | valid | — |
| bedrock → bedrock | cost_window_rps | 5000 | valid | — |
| bedrock → bedrock | cost_threads | 143 | valid | — |
| bedrock → bedrock | cost_nonvol_ctxt_per_request | 0.005065484720231946 | valid | — |
| bedrock → bedrock | cost_majflt | 0 | valid | — |
| bedrock → bedrock | added_ttft_p50_us | untestable | untestable | — |
| bedrock → bedrock | added_ttft_p99_us | untestable | untestable | — |
| bedrock → bedrock | added_gap_p50_us | untestable | untestable | — |
| bedrock → bedrock | added_gap_p99_us | untestable | untestable | — |
| bedrock → bedrock | streams_sustained_fps | untestable | untestable | — |
| bedrock → bedrock | streams_sustained | untestable | untestable | — |
| bedrock → bedrock | idle_rss_mib | 1157.3984375 | valid | — |
| bedrock → bedrock | steady_state_rss_mib | 1165.7421875 | valid | — |
| bedrock → bedrock | recovered_rss_mib | 1167.375 | valid | — |
| bedrock → bedrock | peak_rss_mib | 1175.8359375 | valid | — |
| bedrock → bedrock | peak_rss_hwm_mib | 1275.18359375 | valid | — |
| bedrock → bedrock | time_to_plateau_s | 60 | valid | — |
| bedrock → bedrock | growth_rate_mib_per_min | 0.6896154469149554 | valid | — |
| bedrock → bedrock | load_s | 300 | valid | — |
| bedrock → bedrock | shape | not_measured | not_measured | — |
| bedrock → bedrock | idle_shape | not_measured | not_measured | — |
| bedrock → bedrock | idle_static | 1 | valid | — |
| bedrock → bedrock | idle_growth_rate_mib_per_min | 0.6918641708843657 | valid | — |
| bedrock → bedrock | 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 | throughput_rps | 4996 req/s | concurrency 8; samples 29985 | 5 |
| bedrock → bedrock | throughput_rps | 5179 req/s | concurrency 16; samples 31083 | 10 |
| bedrock → bedrock | throughput_rps | 5403 req/s | concurrency 64; samples 32432 | 50 |
| bedrock → bedrock | throughput_rps | 5403 req/s | concurrency 64; samples 32432 | 100 |
| bedrock → bedrock | throughput_rps | 5403 req/s | concurrency 64; samples 32432 | unbounded |
| openai → bedrock | added_latency_p50_us | 929 | valid; gateway_c1_samples=6162; direct_c1_samples=262294 | — |
| openai → bedrock | added_latency_p99_us | 1235 | valid; gateway_c1_samples=6162; direct_c1_samples=262294 | — |
| openai → bedrock | gateway_c1_p99_us | 1263 | valid; gateway_c1_samples=6162 | — |
| openai → bedrock | direct_c1_p99_us | 28 | valid; direct_c1_samples=262294 | — |
| openai → bedrock | cpu_us_per_request | 912.3461771232601 | valid | — |
| openai → bedrock | rps_per_cpu_second | 1096.0751796572692 | valid | — |
| openai → bedrock | cost_window_conc | 8 | valid | — |
| openai → bedrock | cost_core_utilisation | 0.7513336009437701 | valid | — |
| openai → bedrock | cost_window_ok | 19828 | valid | — |
| openai → bedrock | cost_window_rps | 3304 | valid | — |
| openai → bedrock | cost_threads | 144 | valid | — |
| openai → bedrock | cost_nonvol_ctxt_per_request | 0.008069396812588258 | valid | — |
| openai → bedrock | cost_majflt | 0 | valid | — |
| openai → bedrock | added_ttft_p50_us | untestable | untestable | — |
| openai → bedrock | added_ttft_p99_us | untestable | untestable | — |
| openai → bedrock | added_gap_p50_us | untestable | untestable | — |
| openai → bedrock | added_gap_p99_us | untestable | untestable | — |
| openai → bedrock | streams_sustained_fps | untestable | untestable | — |
| openai → bedrock | streams_sustained | untestable | untestable | — |
| openai → bedrock | idle_rss_mib | 1158.6875 | valid | — |
| openai → bedrock | steady_state_rss_mib | 1172.09765625 | valid | — |
| openai → bedrock | recovered_rss_mib | 1173.9765625 | valid | — |
| openai → bedrock | peak_rss_mib | 1179.87109375 | valid | — |
| openai → bedrock | peak_rss_hwm_mib | 1198.6875 | valid | — |
| openai → bedrock | time_to_plateau_s | 60 | valid | — |
| openai → bedrock | growth_rate_mib_per_min | 0.9309596103361604 | valid | — |
| openai → bedrock | load_s | 300 | valid | — |
| openai → bedrock | shape | not_measured | not_measured | — |
| openai → bedrock | idle_shape | not_measured | not_measured | — |
| openai → bedrock | idle_static | 1 | valid | — |
| openai → bedrock | idle_growth_rate_mib_per_min | 1.0752406083257144 | valid | — |
| openai → bedrock | 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 | throughput_rps | 4316 req/s | concurrency 8; samples 25900 | 5 |
| openai → bedrock | throughput_rps | 4440 req/s | concurrency 16; samples 26646 | 10 |
| openai → bedrock | throughput_rps | 4587 req/s | concurrency 64; samples 27530 | 50 |
| openai → bedrock | throughput_rps | 4587 req/s | concurrency 64; samples 27530 | 100 |
| openai → bedrock | throughput_rps | 4587 req/s | concurrency 64; samples 27530 | unbounded |
| anthropic → bedrock | — | — | not_declared: Higress 2.2.5 with bundled AI Proxy 2.0.3. Source 4deae1dd3b4cb2707273748985e55bef28e747d1 and pinned-image probes establish 13 configured exact-wire pairs; two Cohere translations failed response correctness in the preflight. In the 2026-11-09 run the engine withheld performance and streaming numbers for six configured pairs because re-verification found the request did not arrive on the requested upstream wire: OpenAI to Anthropic, Gemini, Cohere and Bedrock arrived untranslated on the mock's OpenAI Chat Completions endpoint, and Anthropic to Gemini and Anthropic to Cohere also arrived on the OpenAI Chat Completions endpoint. Responses cross-conversion is not implemented; Messages-to-Bedrock uses Anthropic Mantle rather than Converse. Pairs outside the configured 13 are not configured: unsupported translations pass through the original wire or return unsupported API errors. Evidence: results/diagnostics/refresh-configuration/higress-adapter-preflight.json | — |
| openai → openai | added_latency_p50_us | 718 | valid; gateway_c1_samples=7996; direct_c1_samples=262593 | — |
| openai → openai | added_latency_p99_us | 943 | valid; gateway_c1_samples=7996; direct_c1_samples=262593 | — |
| openai → openai | gateway_c1_p99_us | 971 | valid; gateway_c1_samples=7996 | — |
| openai → openai | direct_c1_p99_us | 28 | valid; direct_c1_samples=262593 | — |
| openai → openai | cpu_us_per_request | 709.3196158019531 | valid | — |
| openai → openai | rps_per_cpu_second | 1409.801699716714 | valid | — |
| openai → openai | cost_window_conc | 8 | valid | — |
| openai → openai | cost_core_utilisation | 0.7330563381485045 | valid | — |
| openai → openai | cost_window_ok | 24883 | valid | — |
| openai → openai | cost_window_rps | 4146 | valid | — |
| openai → openai | cost_threads | 142 | valid | — |
| openai → openai | cost_nonvol_ctxt_per_request | 0.008399308765020295 | valid | — |
| openai → openai | cost_majflt | 0 | valid | — |
| openai → openai | added_ttft_p50_us | 881 | valid; ttft_gw_samples=100; ttft_direct_samples=100 | — |
| openai → openai | added_ttft_p99_us | 1103 | valid; ttft_gw_samples=100; ttft_direct_samples=100 | — |
| openai → openai | added_gap_p50_us | 2 | 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 | 52911.767627714544 | valid | — |
| openai → openai | streams_sustained | 1688 | valid | — |
| openai → openai | idle_rss_mib | 1163.541015625 | valid | — |
| openai → openai | steady_state_rss_mib | 1170.94140625 | valid | — |
| openai → openai | recovered_rss_mib | 1171.9921875 | valid | — |
| openai → openai | peak_rss_mib | 1181.8125 | valid | — |
| openai → openai | peak_rss_hwm_mib | 1287.14453125 | valid | — |
| openai → openai | time_to_plateau_s | 66 | valid | — |
| openai → openai | growth_rate_mib_per_min | 0.8963186649029332 | valid | — |
| openai → openai | load_s | 300 | valid | — |
| openai → openai | shape | not_measured | not_measured | — |
| openai → openai | idle_shape | not_measured | not_measured | — |
| openai → openai | idle_static | 1 | valid | — |
| openai → openai | idle_growth_rate_mib_per_min | 0.6061112080903336 | valid | — |
| openai → openai | throughput_rps | 2688 req/s | concurrency 2; samples 16131 | 1 |
| openai → openai | throughput_rps | 5685 req/s | concurrency 8; samples 34120 | 5 |
| openai → openai | throughput_rps | 6095 req/s | concurrency 32; samples 36576 | 10 |
| openai → openai | throughput_rps | 6138 req/s | concurrency 64; samples 36832 | 50 |
| openai → openai | throughput_rps | 6138 req/s | concurrency 64; samples 36832 | 100 |
| openai → openai | throughput_rps | 6138 req/s | concurrency 64; samples 36832 | unbounded |
| gemini → openai | — | — | not_declared: Higress 2.2.5 with bundled AI Proxy 2.0.3. Source 4deae1dd3b4cb2707273748985e55bef28e747d1 and pinned-image probes establish 13 configured exact-wire pairs; two Cohere translations failed response correctness in the preflight. In the 2026-11-09 run the engine withheld performance and streaming numbers for six configured pairs because re-verification found the request did not arrive on the requested upstream wire: OpenAI to Anthropic, Gemini, Cohere and Bedrock arrived untranslated on the mock's OpenAI Chat Completions endpoint, and Anthropic to Gemini and Anthropic to Cohere also arrived on the OpenAI Chat Completions endpoint. Responses cross-conversion is not implemented; Messages-to-Bedrock uses Anthropic Mantle rather than Converse. Pairs outside the configured 13 are not configured: unsupported translations pass through the original wire or return unsupported API errors. Evidence: results/diagnostics/refresh-configuration/higress-adapter-preflight.json | — |
| bedrock → openai | — | — | not_declared: Higress 2.2.5 with bundled AI Proxy 2.0.3. Source 4deae1dd3b4cb2707273748985e55bef28e747d1 and pinned-image probes establish 13 configured exact-wire pairs; two Cohere translations failed response correctness in the preflight. In the 2026-11-09 run the engine withheld performance and streaming numbers for six configured pairs because re-verification found the request did not arrive on the requested upstream wire: OpenAI to Anthropic, Gemini, Cohere and Bedrock arrived untranslated on the mock's OpenAI Chat Completions endpoint, and Anthropic to Gemini and Anthropic to Cohere also arrived on the OpenAI Chat Completions endpoint. Responses cross-conversion is not implemented; Messages-to-Bedrock uses Anthropic Mantle rather than Converse. Pairs outside the configured 13 are not configured: unsupported translations pass through the original wire or return unsupported API errors. Evidence: results/diagnostics/refresh-configuration/higress-adapter-preflight.json | — |
| cohere → openai | — | — | not_declared: Higress 2.2.5 with bundled AI Proxy 2.0.3. Source 4deae1dd3b4cb2707273748985e55bef28e747d1 and pinned-image probes establish 13 configured exact-wire pairs; two Cohere translations failed response correctness in the preflight. In the 2026-11-09 run the engine withheld performance and streaming numbers for six configured pairs because re-verification found the request did not arrive on the requested upstream wire: OpenAI to Anthropic, Gemini, Cohere and Bedrock arrived untranslated on the mock's OpenAI Chat Completions endpoint, and Anthropic to Gemini and Anthropic to Cohere also arrived on the OpenAI Chat Completions endpoint. Responses cross-conversion is not implemented; Messages-to-Bedrock uses Anthropic Mantle rather than Converse. Pairs outside the configured 13 are not configured: unsupported translations pass through the original wire or return unsupported API errors. Evidence: results/diagnostics/refresh-configuration/higress-adapter-preflight.json | — |
| anthropic → openai | added_latency_p50_us | 954 | valid; gateway_c1_samples=6058; direct_c1_samples=259722 | — |
| anthropic → openai | added_latency_p99_us | 1252 | valid; gateway_c1_samples=6058; direct_c1_samples=259722 | — |
| anthropic → openai | gateway_c1_p99_us | 1280 | valid; gateway_c1_samples=6058 | — |
| anthropic → openai | direct_c1_p99_us | 28 | valid; direct_c1_samples=259722 | — |
| anthropic → openai | cpu_us_per_request | 968.311271560369 | valid | — |
| anthropic → openai | rps_per_cpu_second | 1032.7257663628832 | valid | — |
| anthropic → openai | cost_window_conc | 8 | valid | — |
| anthropic → openai | cost_core_utilisation | 0.5011834686494262 | valid | — |
| anthropic → openai | cost_window_ok | 12465 | valid | — |
| anthropic → openai | cost_window_rps | 2076 | valid | — |
| anthropic → openai | cost_threads | 142 | valid | — |
| anthropic → openai | cost_nonvol_ctxt_per_request | 0.00008022462896109105 | valid | — |
| anthropic → openai | cost_majflt | 0 | valid | — |
| anthropic → openai | added_ttft_p50_us | 1262 | valid; ttft_gw_samples=100; ttft_direct_samples=100 | — |
| anthropic → openai | added_ttft_p99_us | 1705 | valid; ttft_gw_samples=100; ttft_direct_samples=100 | — |
| anthropic → openai | added_gap_p50_us | 8 | valid; gap_gw_samples=63; gap_direct_samples=63 | — |
| anthropic → openai | added_gap_p99_us | 163 | valid; gap_gw_samples=63; gap_direct_samples=63 | — |
| anthropic → openai | streams_sustained_fps | 19376.729819513093 | valid | — |
| anthropic → openai | streams_sustained | 5088 | valid | — |
| anthropic → openai | idle_rss_mib | 1151.65234375 | valid | — |
| anthropic → openai | steady_state_rss_mib | 1160.24609375 | valid | — |
| anthropic → openai | recovered_rss_mib | 1161.1953125 | valid | — |
| anthropic → openai | peak_rss_mib | 1177.22265625 | valid | — |
| anthropic → openai | peak_rss_hwm_mib | 1280.44921875 | valid | — |
| anthropic → openai | time_to_plateau_s | 60 | valid | — |
| anthropic → openai | growth_rate_mib_per_min | 0.5290055438736946 | valid | — |
| anthropic → openai | load_s | 300 | valid | — |
| anthropic → openai | shape | not_measured | not_measured | — |
| anthropic → openai | idle_shape | not_measured | not_measured | — |
| anthropic → openai | idle_static | 1 | valid | — |
| anthropic → openai | idle_growth_rate_mib_per_min | 1.003040818815334 | valid | — |
| anthropic → 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 |
| anthropic → openai | throughput_rps | 3867 req/s | concurrency 8; samples 23211 | 5 |
| anthropic → openai | throughput_rps | 3999 req/s | concurrency 16; samples 23998 | 10 |
| anthropic → openai | throughput_rps | 4118 req/s | concurrency 64; samples 24713 | 50 |
| anthropic → openai | throughput_rps | 4118 req/s | concurrency 64; samples 24713 | 100 |
| anthropic → openai | throughput_rps | 4118 req/s | concurrency 64; samples 24713 | unbounded |
| openai-responses → openai | — | — | not_declared: Higress 2.2.5 with bundled AI Proxy 2.0.3. Source 4deae1dd3b4cb2707273748985e55bef28e747d1 and pinned-image probes establish 13 configured exact-wire pairs; two Cohere translations failed response correctness in the preflight. In the 2026-11-09 run the engine withheld performance and streaming numbers for six configured pairs because re-verification found the request did not arrive on the requested upstream wire: OpenAI to Anthropic, Gemini, Cohere and Bedrock arrived untranslated on the mock's OpenAI Chat Completions endpoint, and Anthropic to Gemini and Anthropic to Cohere also arrived on the OpenAI Chat Completions endpoint. Responses cross-conversion is not implemented; Messages-to-Bedrock uses Anthropic Mantle rather than Converse. Pairs outside the configured 13 are not configured: unsupported translations pass through the original wire or return unsupported API errors. Evidence: results/diagnostics/refresh-configuration/higress-adapter-preflight.json | — |

## 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 |
| --- | --- | --- | --- | --- | --- | --- | --- |
| cohere → cohere | served | — | — | — | Recorded run evidence | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| bedrock → cohere | 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) | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| gemini → cohere | 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) | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| openai → cohere | not_served | 200 | failed | probed and answered HTTP 200, but the body is not a usable openai response (see body_snippet) | Recorded run evidence | `the 2xx body is not a response of the openai dialect with assistant text at choices[].message.content: {"response_id":"x","generation_id":"x","text":"ok","finish_reason":"COMPLETE","meta":{"api_version":{"version":"1"},"tokens":{"input_tokens":10,"output_tokens":2},"billed_units":{"input_tokens":10,"output_tokens":2}}}` | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| anthropic → cohere | not_served | 200 | failed | probed and answered HTTP 200, but the body is not a usable anthropic response (see body_snippet) | Recorded run evidence | `the 2xx body is not a response of the anthropic dialect with assistant text at content[type=text].text: {"id":"","type":"message","role":"assistant","content":null,"model":"","stop_reason":null,"stop_sequence":null,"usage":{},"error":null}` | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| openai-responses → cohere | 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) | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| cohere → anthropic | 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) | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| gemini → anthropic | 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) | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| openai → anthropic | served | — | — | — | Recorded run evidence | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| bedrock → anthropic | 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) | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| openai-responses → anthropic | 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) | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| anthropic → anthropic | served | — | — | — | Recorded run evidence | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| 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) | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| gemini → 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) | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| 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) | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| bedrock → 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) | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| openai-responses → openai-responses | served | — | — | — | Recorded run evidence | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| anthropic → 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) | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| openai → gemini | served | — | — | — | Recorded run evidence | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| gemini → gemini | served | — | — | — | Recorded run evidence | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| bedrock → gemini | 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) | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| cohere → gemini | 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) | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| openai-responses → gemini | 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) | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| anthropic → gemini | served | — | — | — | Recorded run evidence | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| openai-responses → bedrock | 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) | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| gemini → bedrock | 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) | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| cohere → bedrock | 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) | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| bedrock → bedrock | served | — | — | — | Recorded run evidence | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| openai → bedrock | served | — | — | — | Recorded run evidence | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| anthropic → bedrock | 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) | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| openai → openai | served | — | — | — | Recorded run evidence | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| gemini → openai | 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) | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| bedrock → openai | 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) | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| cohere → openai | 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) | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| anthropic → openai | served | — | — | — | Recorded run evidence | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |
| openai-responses → openai | 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) | — | higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422 |

**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 → 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 | 2048 | {"200":2048} | 0 |
| openai → anthropic | 4096 | {"200":4096} | 0 |
| openai → anthropic | 8192 | {"200":8082} | 110 |
| openai → anthropic | 6144 | {"200":6144} | 0 |
| openai → anthropic | 7168 | {"200":7168} | 0 |
| openai → anthropic | 7680 | {"200":7680} | 0 |
| openai → anthropic | 7424 | {"200":7424} | 0 |
| openai → anthropic | 7552 | {"200":7552} | 0 |
| openai → anthropic | 7488 | {"200":7488} | 0 |
| openai → anthropic | 7456 | {"200":7456} | 0 |
| openai → anthropic | 7440 | {"200":7440} | 0 |
| openai → anthropic | 7432 | {"200":7432} | 0 |
| openai → anthropic | 7436 | {"200":7436} | 0 |
| openai → anthropic | 7438 | {"200":7438} | 0 |
| openai → anthropic | 7439 | {"200":7439} | 0 |
| openai → anthropic | 7438 | {"200":7438} | 0 |
| openai → anthropic | 7438 | {"200":7438} | 0 |
| openai → anthropic | 5767 | {"200":5767} | 0 |
| openai → anthropic | 5767 | {"200":5767} | 0 |
| openai → anthropic | 5767 | {"200":5767} | 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 | 2048 | {"200":2048} | 0 |
| anthropic → anthropic | 4096 | {"200":4096} | 0 |
| anthropic → anthropic | 8192 | {"200":8192} | 0 |
| anthropic → anthropic | 16384 | {"200":16384} | 0 |
| anthropic → anthropic | 32768 | {"200":17771} | 14997 |
| anthropic → anthropic | 24576 | {"200":17562} | 7014 |
| anthropic → anthropic | 20480 | {"200":17305} | 3175 |
| anthropic → anthropic | 18432 | {"200":17810} | 622 |
| anthropic → anthropic | 17408 | {"200":17408} | 0 |
| anthropic → anthropic | 16896 | {"200":16896} | 0 |
| anthropic → anthropic | 16640 | {"200":16640} | 0 |
| anthropic → anthropic | 16512 | {"200":16512} | 0 |
| anthropic → anthropic | 16448 | {"200":16448} | 0 |
| anthropic → anthropic | 16416 | {"200":16416} | 0 |
| anthropic → anthropic | 16400 | {"200":16400} | 0 |
| anthropic → anthropic | 16392 | {"200":16392} | 0 |
| anthropic → anthropic | 16388 | {"200":16388} | 0 |
| anthropic → anthropic | 16390 | {"200":16390} | 0 |
| anthropic → anthropic | 16391 | {"200":16391} | 0 |
| anthropic → anthropic | 16391 | {"200":16391} | 0 |
| anthropic → anthropic | 16391 | {"200":16391} | 0 |
| 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 | 2048 | {"200":2048} | 0 |
| openai → openai | 1536 | {"200":1536} | 0 |
| openai → openai | 1792 | {"200":1792} | 0 |
| openai → openai | 1664 | {"200":1664} | 0 |
| openai → openai | 1728 | {"200":1728} | 0 |
| openai → openai | 1696 | {"200":1696} | 0 |
| openai → openai | 1680 | {"200":1680} | 0 |
| openai → openai | 1688 | {"200":1688} | 0 |
| openai → openai | 1692 | {"200":1692} | 0 |
| openai → openai | 1690 | {"200":1690} | 0 |
| openai → openai | 1689 | {"200":1689} | 0 |
| openai → openai | 1688 | {"200":1688} | 0 |
| openai → openai | 1688 | {"200":1688} | 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 | 2048 | {"200":2048} | 0 |
| anthropic → openai | 4096 | {"200":4096} | 0 |
| anthropic → openai | 8192 | {"200":6688} | 1504 |
| anthropic → openai | 6144 | {"200":6144} | 0 |
| anthropic → openai | 5120 | {"200":5120} | 0 |
| anthropic → openai | 5632 | {"200":5632} | 0 |
| anthropic → openai | 5888 | {"200":5888} | 0 |
| anthropic → openai | 6016 | {"200":6016} | 0 |
| anthropic → openai | 6080 | {"200":6080} | 0 |
| anthropic → openai | 6112 | {"200":6112} | 0 |
| anthropic → openai | 6096 | {"200":6096} | 0 |
| anthropic → openai | 6088 | {"200":6088} | 0 |
| anthropic → openai | 6084 | {"200":6084} | 0 |
| anthropic → openai | 6082 | {"200":6082} | 0 |
| anthropic → openai | 6081 | {"200":6081} | 0 |
| anthropic → openai | 6080 | {"200":6080} | 0 |
| anthropic → openai | 6080 | {"200":6080} | 0 |
| anthropic → openai | 5088 | {"200":5088} | 0 |
| anthropic → openai | 5088 | {"200":5088} | 0 |
| anthropic → openai | 5088 | {"200":5088} | 0 |

## Recorded notes

Text recorded with the cells above. Each note is shown once; rows refer to it by number.

- Higress 2.2.5 with bundled AI Proxy 2.0.3. Source 4deae1dd3b4cb2707273748985e55bef28e747d1 and pinned-image probes establish 13 configured exact-wire pairs; two Cohere translations failed response correctness in the preflight. In the 2026-11-09 run the engine withheld performance and streaming numbers for six configured pairs because re-verification found the request did not arrive on the requested upstream wire: OpenAI to Anthropic, Gemini, Cohere and Bedrock arrived untranslated on the mock's OpenAI Chat Completions endpoint, and Anthropic to Gemini and Anthropic to Cohere also arrived on the OpenAI Chat Completions endpoint. Responses cross-conversion is not implemented; Messages-to-Bedrock uses Anthropic Mantle rather than Converse. Pairs outside the configured 13 are not configured: unsupported translations pass through the original wire or return unsupported API errors. Evidence: results/diagnostics/refresh-configuration/higress-adapter-preflight.json
- 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-higress-558a1502d4a7ea1f2b74c745739723084c953a3ae5dcf16760a2694a92c21270](https://onthebench.ai/gateways/runs/run-higress-558a1502d4a7ea1f2b74c745739723084c953a3ae5dcf16760a2694a92c21270)  ·  [complete raw record including RSS series](https://onthebench.ai/api/v1/runs/run-higress-558a1502d4a7ea1f2b74c745739723084c953a3ae5dcf16760a2694a92c21270/raw.json)

## Configuration recorded by the run

```
# data/ingresses/anthropic.gen.yaml
apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: anthropic
  namespace: higress-system
  annotations:
    higress.io/destination: mock.static
spec:
  rules:
  - host: anthropic.bench
    http:
      paths:
      - path: /
        pathType: Prefix
        backend:
          resource:
            apiGroup: networking.higress.io
            kind: McpBridge
            name: default

# data/ingresses/openai-responses.gen.yaml
apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: openai-responses
  namespace: higress-system
  annotations:
    higress.io/destination: mock.static
spec:
  rules:
  - host: openai-responses.bench
    http:
      paths:
      - path: /
        pathType: Prefix
        backend:
          resource:
            apiGroup: networking.higress.io
            kind: McpBridge
            name: default

# data/ingresses/bedrock.gen.yaml
apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: bedrock
  namespace: higress-system
  annotations:
    higress.io/destination: mock.static
spec:
  rules:
  - host: bedrock.bench
    http:
      paths:
      - path: /
        pathType: Prefix
        backend:
          resource:
            apiGroup: networking.higress.io
            kind: McpBridge
            name: default

# data/ingresses/gemini.gen.yaml
apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: gemini
  namespace: higress-system
  annotations:
    higress.io/destination: mock.static
spec:
  rules:
  - host: gemini.bench
    http:
      paths:
      - path: /
        pathType: Prefix
        backend:
          resource:
            apiGroup: networking.higress.io
            kind: McpBridge
            name: default

# data/ingresses/openai.gen.yaml
apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: openai
  namespace: higress-system
  annotations:
    higress.io/destination: mock.static
spec:
  rules:
  - http:
      paths:
      - path: /
        pathType: Prefix
        backend:
          resource:
            apiGroup: networking.higress.io
            kind: McpBridge
            name: default

# data/ingresses/cohere.gen.yaml
apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: cohere
  namespace: higress-system
  annotations:
    higress.io/destination: mock.static
spec:
  rules:
  - host: cohere.bench
    http:
      paths:
      - path: /
        pathType: Prefix
        backend:
          resource:
            apiGroup: networking.higress.io
            kind: McpBridge
            name: default

# data/wasmplugins/bench.gen.yaml
apiVersion: extensions.higress.io/v1alpha1
kind: WasmPlugin
metadata:
  name: bench
  namespace: higress-system
spec:
  defaultConfigDisable: true
  matchRules:
  - ingress: [openai]
    config:
      provider:
        type: openai
        apiTokens: [dummy]
        modelMapping:
          "*": gpt-4o-mini
  - ingress: [openai-responses]
    config:
      provider:
        type: openai
        apiTokens: [dummy]
        modelMapping:
          "*": gpt-4o-mini
  - ingress: [anthropic]
    config:
      provider:
        type: claude
        apiTokens: [dummy]
        modelMapping:
          "*": claude-3-5-sonnet-20241022
  - ingress: [gemini]
    config:
      provider:
        type: gemini
        apiTokens: [dummy]
        modelMapping:
          "*": gemini-1.5-pro
  - ingress: [cohere]
    config:
      provider:
        type: cohere
        apiTokens: [dummy]
        modelMapping:
          "*": command-r-plus
  - ingress: [bedrock]
    config:
      provider:
        type: bedrock
        apiTokens: [dummy]
        modelMapping:
          "*": anthropic.claude-3-sonnet-20240229-v1:0
        awsRegion: us-east-1
  url: file:///usr/share/nginx/html/plugins/ai-proxy/2.0.3/plugin.wasm

# data/mcpbridges/default.gen.yaml
# The upstream registry this run drives: the harness's mock, on the host loopback. This file
# replaces the image's own default McpBridge (whose only registry is the console at 8001); the
# all-in-one bootstrap copies its defaults into /data with cp -n, so a file that already exists at
# this path is left alone and this one wins. 127.0.0.1 reaches the mock because the harness
# launches every container with host networking - the same fact apisix's override.endpoint and
# kong's upstream_url rely on. Shape mirrors the image's own all-in-one/config/mcpbridges/default.yaml.
apiVersion: networking.higress.io/v1
kind: McpBridge
metadata:
  name: default
  namespace: higress-system
  creationTimestamp: "2000-01-01T00:00:00Z"
  resourceVersion: "1"
spec:
  registries:
  - domain: "127.0.0.1:8000"
    name: mock
    port: 8000
    type: static

```

## Compare a shared protocol cell

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

## Historical measurements

- [2026-11-10T17:39:42Z — higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5@sha256:9846006cb3cf2dafa942deab0e1d004ed30e219fbc9580438f9e3c58e7096422](https://onthebench.ai/gateways/runs/run-higress-558a1502d4a7ea1f2b74c745739723084c953a3ae5dcf16760a2694a92c21270)
- [2026-08-26T19:01:18Z — higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.4](https://onthebench.ai/gateways/runs/run-higress-af1ce2bee0833f93ed558ceb54c4e30f0164de90a7a837817aa387b275cc353b)
