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.

Other · recorded build: higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5. Measured: .

JSON · Markdown · CSV · Project source · Interactive dashboard

Each row below names its protocol pair. A representative dashboard cell may differ between gateways; compare an identical pair before interpreting a ranking. Unavailable metrics retain their reason; below-resolution readings are not exact zeroes.

Published measurements

Ingress → upstreamMetricValueState / reasonp99 bound (ms)
Cohere → CohereAdded latency p50 (µs)615Valid · n=9,341 gateway / 267,984 direct—
Cohere → CohereAdded latency p99 (µs)747Valid · n=9,341 gateway / 267,984 direct—
Cohere → CohereGateway p99 at concurrency 1 (µs)774Valid · n=9,341—
Cohere → CohereDirect-to-mock p99 at concurrency 1 (µs)27Valid · n=267,984—
Cohere → CohereCPU per request (µs)613.2Valid—
Cohere → CohereRequests per CPU-second1,630.7Valid—
Cohere → CohereCost window concurrency8Valid—
Cohere → CohereCost window core utilisation0.805Valid—
Cohere → CohereCost window successful requests31,620Valid—
Cohere → CohereCost window throughput (req/s)5,269Valid—
Cohere → CohereCost window threads143Valid—
Cohere → CohereNon-voluntary context switches per request0.008Valid—
Cohere → CohereCost window major page faults0Valid—
Cohere → CohereAdded time to first event p50 (µs)—Untestable—
Cohere → CohereAdded time to first event p99 (µs)—Untestable—
Cohere → CohereAdded inter-frame gap p50 (µs)—Untestable—
Cohere → CohereAdded inter-frame gap p99 (µs)—Untestable—
Cohere → CohereStreams sustained (frames/s)—Untestable—
Cohere → CohereStreams sustained—Untestable—
Cohere → CohereIdle RSS (MiB)1,207.9Valid—
Cohere → CohereSteady-state RSS (MiB)1,217.3Valid—
Cohere → CohereRecovered RSS (MiB)1,218.4Valid—
Cohere → CoherePeak RSS (MiB)1,226.8Valid—
Cohere → CoherePeak RSS high-water mark (MiB)1,324.4Valid—
Cohere → CohereTime to plateau (s)60Valid—
Cohere → CohereMemory growth (MiB/min)0.748Valid—
Cohere → CohereMemory load duration (s)300Valid—
Cohere → CohereMemory shape under load—Not measured—
Cohere → CohereIdle memory shape—Not measured—
Cohere → CohereIdle memory static1Valid—
Cohere → CohereIdle memory growth (MiB/min)1.03Valid—
Cohere → CohereThroughput (req/s)3,092concurrency 2; n=18,5551 ms
Cohere → CohereThroughput (req/s)5,921concurrency 8; n=35,5365 ms
Cohere → CohereThroughput (req/s)6,529concurrency 32; n=39,18910 ms
Cohere → CohereThroughput (req/s)6,633concurrency 64; n=39,80550 ms
Cohere → CohereThroughput (req/s)6,633concurrency 64; n=39,805100 ms
Cohere → CohereThroughput (req/s)6,633concurrency 64; n=39,805Unbounded
Bedrock → Cohere——Not declared: recorded note 1—
Gemini → Cohere——Not declared: recorded note 1—
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: recorded note 1—
Cohere → Anthropic——Not declared: recorded note 1—
Gemini → Anthropic——Not declared: recorded note 1—
OpenAI → AnthropicAdded latency p50 (µs)859Valid · n=6,739 gateway / 262,868 direct—
OpenAI → AnthropicAdded latency p99 (µs)1,112Valid · n=6,739 gateway / 262,868 direct—
OpenAI → AnthropicGateway p99 at concurrency 1 (µs)1,140Valid · n=6,739—
OpenAI → AnthropicDirect-to-mock p99 at concurrency 1 (µs)28Valid · n=262,868—
OpenAI → AnthropicCPU per request (µs)882.4Valid—
OpenAI → AnthropicRequests per CPU-second1,133.2Valid—
OpenAI → AnthropicCost window concurrency8Valid—
OpenAI → AnthropicCost window core utilisation0.98Valid—
OpenAI → AnthropicCost window successful requests26,744Valid—
OpenAI → AnthropicCost window throughput (req/s)4,456Valid—
OpenAI → AnthropicCost window threads145Valid—
OpenAI → AnthropicNon-voluntary context switches per request0.001Valid—
OpenAI → AnthropicCost window major page faults0Valid—
OpenAI → AnthropicAdded time to first event p50 (µs)1,095Valid · n=100 gateway / 100 direct—
OpenAI → AnthropicAdded time to first event p99 (µs)1,556Valid · n=100 gateway / 100 direct—
OpenAI → AnthropicAdded inter-frame gap p50 (µs)8Valid · n=63 gateway / 63 direct—
OpenAI → AnthropicAdded inter-frame gap p99 (µs)—Below resolution · n=63 gateway / 63 direct—
OpenAI → AnthropicStreams sustained (frames/s)23,656.6Valid—
OpenAI → AnthropicStreams sustained5,767Valid—
OpenAI → AnthropicIdle RSS (MiB)1,148.4Valid—
OpenAI → AnthropicSteady-state RSS (MiB)1,158.9Valid—
OpenAI → AnthropicRecovered RSS (MiB)1,159.8Valid—
OpenAI → AnthropicPeak RSS (MiB)1,167.3Valid—
OpenAI → AnthropicPeak RSS high-water mark (MiB)1,264.9Valid—
OpenAI → AnthropicTime to plateau (s)60Valid—
OpenAI → AnthropicMemory growth (MiB/min)0.702Valid—
OpenAI → AnthropicMemory load duration (s)300Valid—
OpenAI → AnthropicMemory shape under load—Not measured—
OpenAI → AnthropicIdle memory shape—Not measured—
OpenAI → AnthropicIdle memory static1Valid—
OpenAI → AnthropicIdle memory growth (MiB/min)0.889Valid—
OpenAI → AnthropicThroughput (req/s)—Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound1 ms
OpenAI → AnthropicThroughput (req/s)4,788concurrency 16; n=28,7365 ms
OpenAI → AnthropicThroughput (req/s)4,909concurrency 32; n=29,46410 ms
OpenAI → AnthropicThroughput (req/s)4,944concurrency 64; n=29,67350 ms
OpenAI → AnthropicThroughput (req/s)4,944concurrency 64; n=29,673100 ms
OpenAI → AnthropicThroughput (req/s)4,944concurrency 64; n=29,673Unbounded
Bedrock → Anthropic——Not declared: recorded note 1—
OpenAI Responses → Anthropic——Not declared: recorded note 1—
Anthropic → AnthropicAdded latency p50 (µs)680Valid · n=8,481 gateway / 262,245 direct—
Anthropic → AnthropicAdded latency p99 (µs)822Valid · n=8,481 gateway / 262,245 direct—
Anthropic → AnthropicGateway p99 at concurrency 1 (µs)850Valid · n=8,481—
Anthropic → AnthropicDirect-to-mock p99 at concurrency 1 (µs)28Valid · n=262,245—
Anthropic → AnthropicCPU per request (µs)674.8Valid—
Anthropic → AnthropicRequests per CPU-second1,481.9Valid—
Anthropic → AnthropicCost window concurrency8Valid—
Anthropic → AnthropicCost window core utilisation0.977Valid—
Anthropic → AnthropicCost window successful requests34,868Valid—
Anthropic → AnthropicCost window throughput (req/s)5,810Valid—
Anthropic → AnthropicCost window threads144Valid—
Anthropic → AnthropicNon-voluntary context switches per request0.002Valid—
Anthropic → AnthropicCost window major page faults0Valid—
Anthropic → AnthropicAdded time to first event p50 (µs)822Valid · n=100 gateway / 100 direct—
Anthropic → AnthropicAdded time to first event p99 (µs)1,012Valid · n=100 gateway / 100 direct—
Anthropic → AnthropicAdded inter-frame gap p50 (µs)0Valid · n=63 gateway / 63 direct—
Anthropic → AnthropicAdded inter-frame gap p99 (µs)35Valid · n=63 gateway / 63 direct—
Anthropic → AnthropicStreams sustained (frames/s)52,509.7Valid—
Anthropic → AnthropicStreams sustained16,391Valid—
Anthropic → AnthropicIdle RSS (MiB)1,153.9Valid—
Anthropic → AnthropicSteady-state RSS (MiB)1,161.8Valid—
Anthropic → AnthropicRecovered RSS (MiB)1,162.2Valid—
Anthropic → AnthropicPeak RSS (MiB)1,171.8Valid—
Anthropic → AnthropicPeak RSS high-water mark (MiB)1,272.5Valid—
Anthropic → AnthropicTime to plateau (s)60Valid—
Anthropic → AnthropicMemory growth (MiB/min)0.275Valid—
Anthropic → AnthropicMemory load duration (s)300Valid—
Anthropic → AnthropicMemory shape under load—Not measured—
Anthropic → AnthropicIdle memory shape—Not measured—
Anthropic → AnthropicIdle memory static1Valid—
Anthropic → AnthropicIdle memory growth (MiB/min)0.582Valid—
Anthropic → AnthropicThroughput (req/s)1,352concurrency 1; n=8,1151 ms
Anthropic → AnthropicThroughput (req/s)5,967concurrency 16; n=35,8095 ms
Anthropic → AnthropicThroughput (req/s)6,123concurrency 32; n=36,74310 ms
Anthropic → AnthropicThroughput (req/s)6,224concurrency 128; n=37,35250 ms
Anthropic → AnthropicThroughput (req/s)6,224concurrency 128; n=37,352100 ms
Anthropic → AnthropicThroughput (req/s)6,224concurrency 128; n=37,352Unbounded
OpenAI → OpenAI Responses——Not declared: recorded note 1—
Gemini → OpenAI Responses——Not declared: recorded note 1—
Cohere → OpenAI Responses——Not declared: recorded note 1—
Bedrock → OpenAI Responses——Not declared: recorded note 1—
OpenAI Responses → OpenAI ResponsesAdded latency p50 (µs)642Valid · n=8,971 gateway / 260,207 direct—
OpenAI Responses → OpenAI ResponsesAdded latency p99 (µs)747Valid · n=8,971 gateway / 260,207 direct—
OpenAI Responses → OpenAI ResponsesGateway p99 at concurrency 1 (µs)775Valid · n=8,971—
OpenAI Responses → OpenAI ResponsesDirect-to-mock p99 at concurrency 1 (µs)28Valid · n=260,207—
OpenAI Responses → OpenAI ResponsesCPU per request (µs)633.8Valid—
OpenAI Responses → OpenAI ResponsesRequests per CPU-second1,577.8Valid—
OpenAI Responses → OpenAI ResponsesCost window concurrency8Valid—
OpenAI Responses → OpenAI ResponsesCost window core utilisation0.986Valid—
OpenAI Responses → OpenAI ResponsesCost window successful requests37,457Valid—
OpenAI Responses → OpenAI ResponsesCost window throughput (req/s)6,242Valid—
OpenAI Responses → OpenAI ResponsesCost window threads141Valid—
OpenAI Responses → OpenAI ResponsesNon-voluntary context switches per request0.001Valid—
OpenAI Responses → OpenAI ResponsesCost window major page faults0Valid—
OpenAI Responses → OpenAI ResponsesAdded time to first event p50 (µs)—Untestable—
OpenAI Responses → OpenAI ResponsesAdded time to first event p99 (µs)—Untestable—
OpenAI Responses → OpenAI ResponsesAdded inter-frame gap p50 (µs)—Untestable—
OpenAI Responses → OpenAI ResponsesAdded inter-frame gap p99 (µs)—Untestable—
OpenAI Responses → OpenAI ResponsesStreams sustained (frames/s)—Untestable—
OpenAI Responses → OpenAI ResponsesStreams sustained—Untestable—
OpenAI Responses → OpenAI ResponsesIdle RSS (MiB)1,147.2Valid—
OpenAI Responses → OpenAI ResponsesSteady-state RSS (MiB)1,156.6Valid—
OpenAI Responses → OpenAI ResponsesRecovered RSS (MiB)1,157.9Valid—
OpenAI Responses → OpenAI ResponsesPeak RSS (MiB)1,165.5Valid—
OpenAI Responses → OpenAI ResponsesPeak RSS high-water mark (MiB)1,186.6Valid—
OpenAI Responses → OpenAI ResponsesTime to plateau (s)66Valid—
OpenAI Responses → OpenAI ResponsesMemory growth (MiB/min)0.599Valid—
OpenAI Responses → OpenAI ResponsesMemory load duration (s)300Valid—
OpenAI Responses → OpenAI ResponsesMemory shape under load—Not measured—
OpenAI Responses → OpenAI ResponsesIdle memory shape—Not measured—
OpenAI Responses → OpenAI ResponsesIdle memory static1Valid—
OpenAI Responses → OpenAI ResponsesIdle memory growth (MiB/min)0.697Valid—
OpenAI Responses → OpenAI ResponsesThroughput (req/s)2,925concurrency 2; n=17,5531 ms
OpenAI Responses → OpenAI ResponsesThroughput (req/s)6,438concurrency 16; n=38,6365 ms
OpenAI Responses → OpenAI ResponsesThroughput (req/s)6,717concurrency 32; n=40,31110 ms
OpenAI Responses → OpenAI ResponsesThroughput (req/s)6,791concurrency 128; n=40,75450 ms
OpenAI Responses → OpenAI ResponsesThroughput (req/s)6,791concurrency 128; n=40,754100 ms
OpenAI Responses → OpenAI ResponsesThroughput (req/s)6,791concurrency 128; n=40,754Unbounded
Anthropic → OpenAI Responses——Not declared: recorded note 1—
OpenAI → GeminiAdded latency p50 (µs)923Valid · n=6,239 gateway / 265,561 direct—
OpenAI → GeminiAdded latency p99 (µs)1,226Valid · n=6,239 gateway / 265,561 direct—
OpenAI → GeminiGateway p99 at concurrency 1 (µs)1,254Valid · n=6,239—
OpenAI → GeminiDirect-to-mock p99 at concurrency 1 (µs)28Valid · n=265,561—
OpenAI → GeminiCPU per request (µs)951.7Valid—
OpenAI → GeminiRequests per CPU-second1,050.8Valid—
OpenAI → GeminiCost window concurrency8Valid—
OpenAI → GeminiCost window core utilisation0.751Valid—
OpenAI → GeminiCost window successful requests19,009Valid—
OpenAI → GeminiCost window throughput (req/s)3,167Valid—
OpenAI → GeminiCost window threads141Valid—
OpenAI → GeminiNon-voluntary context switches per request0.011Valid—
OpenAI → GeminiCost window major page faults0Valid—
OpenAI → GeminiAdded time to first event p50 (µs)—Untestable—
OpenAI → GeminiAdded time to first event p99 (µs)—Untestable—
OpenAI → GeminiAdded inter-frame gap p50 (µs)—Untestable—
OpenAI → GeminiAdded inter-frame gap p99 (µs)—Untestable—
OpenAI → GeminiStreams sustained (frames/s)—Untestable—
OpenAI → GeminiStreams sustained—Untestable—
OpenAI → GeminiIdle RSS (MiB)1,150.6Valid—
OpenAI → GeminiSteady-state RSS (MiB)1,160.3Valid—
OpenAI → GeminiRecovered RSS (MiB)1,160.6Valid—
OpenAI → GeminiPeak RSS (MiB)1,170.9Valid—
OpenAI → GeminiPeak RSS high-water mark (MiB)1,268.9Valid—
OpenAI → GeminiTime to plateau (s)60Valid—
OpenAI → GeminiMemory growth (MiB/min)0.259Valid—
OpenAI → GeminiMemory load duration (s)300Valid—
OpenAI → GeminiMemory shape under load—Not measured—
OpenAI → GeminiIdle memory shape—Not measured—
OpenAI → GeminiIdle memory static1Valid—
OpenAI → GeminiIdle memory growth (MiB/min)0.605Valid—
OpenAI → GeminiThroughput (req/s)—Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound1 ms
OpenAI → GeminiThroughput (req/s)4,171concurrency 8; n=25,0355 ms
OpenAI → GeminiThroughput (req/s)4,327concurrency 16; n=25,96910 ms
OpenAI → GeminiThroughput (req/s)4,449concurrency 64; n=26,69950 ms
OpenAI → GeminiThroughput (req/s)4,449concurrency 64; n=26,699100 ms
OpenAI → GeminiThroughput (req/s)4,449concurrency 64; n=26,699Unbounded
Gemini → GeminiAdded latency p50 (µs)750Valid · n=7,714 gateway / 265,544 direct—
Gemini → GeminiAdded latency p99 (µs)932Valid · n=7,714 gateway / 265,544 direct—
Gemini → GeminiGateway p99 at concurrency 1 (µs)959Valid · n=7,714—
Gemini → GeminiDirect-to-mock p99 at concurrency 1 (µs)27Valid · n=265,544—
Gemini → GeminiCPU per request (µs)737.7Valid—
Gemini → GeminiRequests per CPU-second1,355.5Valid—
Gemini → GeminiCost window concurrency8Valid—
Gemini → GeminiCost window core utilisation0.743Valid—
Gemini → GeminiCost window successful requests24,237Valid—
Gemini → GeminiCost window throughput (req/s)4,038Valid—
Gemini → GeminiCost window threads140Valid—
Gemini → GeminiNon-voluntary context switches per request0.007Valid—
Gemini → GeminiCost window major page faults0Valid—
Gemini → GeminiAdded time to first event p50 (µs)—Untestable—
Gemini → GeminiAdded time to first event p99 (µs)—Untestable—
Gemini → GeminiAdded inter-frame gap p50 (µs)—Untestable—
Gemini → GeminiAdded inter-frame gap p99 (µs)—Untestable—
Gemini → GeminiStreams sustained (frames/s)—Untestable—
Gemini → GeminiStreams sustained—Untestable—
Gemini → GeminiIdle RSS (MiB)1,156.9Valid—
Gemini → GeminiSteady-state RSS (MiB)1,165.8Valid—
Gemini → GeminiRecovered RSS (MiB)1,166.6Valid—
Gemini → GeminiPeak RSS (MiB)1,175.8Valid—
Gemini → GeminiPeak RSS high-water mark (MiB)1,269.3Valid—
Gemini → GeminiTime to plateau (s)66Valid—
Gemini → GeminiMemory growth (MiB/min)1.08Valid—
Gemini → GeminiMemory load duration (s)300Valid—
Gemini → GeminiMemory shape under load—Not measured—
Gemini → GeminiIdle memory shape—Not measured—
Gemini → GeminiIdle memory static1Valid—
Gemini → GeminiIdle memory growth (MiB/min)0.895Valid—
Gemini → GeminiThroughput (req/s)—Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound1 ms
Gemini → GeminiThroughput (req/s)5,241concurrency 16; n=31,4525 ms
Gemini → GeminiThroughput (req/s)5,428concurrency 32; n=32,57710 ms
Gemini → GeminiThroughput (req/s)5,506concurrency 64; n=33,04550 ms
Gemini → GeminiThroughput (req/s)5,506concurrency 64; n=33,045100 ms
Gemini → GeminiThroughput (req/s)5,506concurrency 64; n=33,045Unbounded
Bedrock → Gemini——Not declared: recorded note 1—
Cohere → Gemini——Not declared: recorded note 1—
OpenAI Responses → Gemini——Not declared: recorded note 1—
Anthropic → GeminiAdded latency p50 (µs)1,045Valid · n=5,488 gateway / 262,925 direct—
Anthropic → GeminiAdded latency p99 (µs)1,360Valid · n=5,488 gateway / 262,925 direct—
Anthropic → GeminiGateway p99 at concurrency 1 (µs)1,388Valid · n=5,488—
Anthropic → GeminiDirect-to-mock p99 at concurrency 1 (µs)28Valid · n=262,925—
Anthropic → GeminiCPU per request (µs)1,055.3Valid—
Anthropic → GeminiRequests per CPU-second947.6Valid—
Anthropic → GeminiCost window concurrency8Valid—
Anthropic → GeminiCost window core utilisation0.744Valid—
Anthropic → GeminiCost window successful requests16,971Valid—
Anthropic → GeminiCost window throughput (req/s)2,828Valid—
Anthropic → GeminiCost window threads142Valid—
Anthropic → GeminiNon-voluntary context switches per request0.01Valid—
Anthropic → GeminiCost window major page faults0Valid—
Anthropic → GeminiAdded time to first event p50 (µs)—Untestable—
Anthropic → GeminiAdded time to first event p99 (µs)—Untestable—
Anthropic → GeminiAdded inter-frame gap p50 (µs)—Untestable—
Anthropic → GeminiAdded inter-frame gap p99 (µs)—Untestable—
Anthropic → GeminiStreams sustained (frames/s)—Untestable—
Anthropic → GeminiStreams sustained—Untestable—
Anthropic → GeminiIdle RSS (MiB)1,151.5Valid—
Anthropic → GeminiSteady-state RSS (MiB)1,164.6Valid—
Anthropic → GeminiRecovered RSS (MiB)1,165.8Valid—
Anthropic → GeminiPeak RSS (MiB)1,173.3Valid—
Anthropic → GeminiPeak RSS high-water mark (MiB)1,280.2Valid—
Anthropic → GeminiTime to plateau (s)60Valid—
Anthropic → GeminiMemory growth (MiB/min)1.17Valid—
Anthropic → GeminiMemory load duration (s)300Valid—
Anthropic → GeminiMemory shape under load—Not measured—
Anthropic → GeminiIdle memory shape—Not measured—
Anthropic → GeminiIdle memory static1Valid—
Anthropic → GeminiIdle memory growth (MiB/min)1.21Valid—
Anthropic → GeminiThroughput (req/s)—Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound1 ms
Anthropic → GeminiThroughput (req/s)3,821concurrency 8; n=22,9295 ms
Anthropic → GeminiThroughput (req/s)3,862concurrency 16; n=23,18510 ms
Anthropic → GeminiThroughput (req/s)3,961concurrency 64; n=23,77150 ms
Anthropic → GeminiThroughput (req/s)3,961concurrency 64; n=23,771100 ms
Anthropic → GeminiThroughput (req/s)3,961concurrency 64; n=23,771Unbounded
OpenAI Responses → Bedrock——Not declared: recorded note 1—
Gemini → Bedrock——Not declared: recorded note 1—
Cohere → Bedrock——Not declared: recorded note 1—
Bedrock → BedrockAdded latency p50 (µs)727Valid · n=7,954 gateway / 263,620 direct—
Bedrock → BedrockAdded latency p99 (µs)861Valid · n=7,954 gateway / 263,620 direct—
Bedrock → BedrockGateway p99 at concurrency 1 (µs)889Valid · n=7,954—
Bedrock → BedrockDirect-to-mock p99 at concurrency 1 (µs)28Valid · n=263,620—
Bedrock → BedrockCPU per request (µs)778.2Valid—
Bedrock → BedrockRequests per CPU-second1,285.1Valid—
Bedrock → BedrockCost window concurrency8Valid—
Bedrock → BedrockCost window core utilisation0.97Valid—
Bedrock → BedrockCost window successful requests30,007Valid—
Bedrock → BedrockCost window throughput (req/s)5,000Valid—
Bedrock → BedrockCost window threads143Valid—
Bedrock → BedrockNon-voluntary context switches per request0.005Valid—
Bedrock → BedrockCost window major page faults0Valid—
Bedrock → BedrockAdded time to first event p50 (µs)—Untestable—
Bedrock → BedrockAdded time to first event p99 (µs)—Untestable—
Bedrock → BedrockAdded inter-frame gap p50 (µs)—Untestable—
Bedrock → BedrockAdded inter-frame gap p99 (µs)—Untestable—
Bedrock → BedrockStreams sustained (frames/s)—Untestable—
Bedrock → BedrockStreams sustained—Untestable—
Bedrock → BedrockIdle RSS (MiB)1,157.4Valid—
Bedrock → BedrockSteady-state RSS (MiB)1,165.7Valid—
Bedrock → BedrockRecovered RSS (MiB)1,167.4Valid—
Bedrock → BedrockPeak RSS (MiB)1,175.8Valid—
Bedrock → BedrockPeak RSS high-water mark (MiB)1,275.2Valid—
Bedrock → BedrockTime to plateau (s)60Valid—
Bedrock → BedrockMemory growth (MiB/min)0.69Valid—
Bedrock → BedrockMemory load duration (s)300Valid—
Bedrock → BedrockMemory shape under load—Not measured—
Bedrock → BedrockIdle memory shape—Not measured—
Bedrock → BedrockIdle memory static1Valid—
Bedrock → BedrockIdle memory growth (MiB/min)0.692Valid—
Bedrock → BedrockThroughput (req/s)—Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound1 ms
Bedrock → BedrockThroughput (req/s)4,996concurrency 8; n=29,9855 ms
Bedrock → BedrockThroughput (req/s)5,179concurrency 16; n=31,08310 ms
Bedrock → BedrockThroughput (req/s)5,403concurrency 64; n=32,43250 ms
Bedrock → BedrockThroughput (req/s)5,403concurrency 64; n=32,432100 ms
Bedrock → BedrockThroughput (req/s)5,403concurrency 64; n=32,432Unbounded
OpenAI → BedrockAdded latency p50 (µs)929Valid · n=6,162 gateway / 262,294 direct—
OpenAI → BedrockAdded latency p99 (µs)1,235Valid · n=6,162 gateway / 262,294 direct—
OpenAI → BedrockGateway p99 at concurrency 1 (µs)1,263Valid · n=6,162—
OpenAI → BedrockDirect-to-mock p99 at concurrency 1 (µs)28Valid · n=262,294—
OpenAI → BedrockCPU per request (µs)912.3Valid—
OpenAI → BedrockRequests per CPU-second1,096.1Valid—
OpenAI → BedrockCost window concurrency8Valid—
OpenAI → BedrockCost window core utilisation0.751Valid—
OpenAI → BedrockCost window successful requests19,828Valid—
OpenAI → BedrockCost window throughput (req/s)3,304Valid—
OpenAI → BedrockCost window threads144Valid—
OpenAI → BedrockNon-voluntary context switches per request0.008Valid—
OpenAI → BedrockCost window major page faults0Valid—
OpenAI → BedrockAdded time to first event p50 (µs)—Untestable—
OpenAI → BedrockAdded time to first event p99 (µs)—Untestable—
OpenAI → BedrockAdded inter-frame gap p50 (µs)—Untestable—
OpenAI → BedrockAdded inter-frame gap p99 (µs)—Untestable—
OpenAI → BedrockStreams sustained (frames/s)—Untestable—
OpenAI → BedrockStreams sustained—Untestable—
OpenAI → BedrockIdle RSS (MiB)1,158.7Valid—
OpenAI → BedrockSteady-state RSS (MiB)1,172.1Valid—
OpenAI → BedrockRecovered RSS (MiB)1,174Valid—
OpenAI → BedrockPeak RSS (MiB)1,179.9Valid—
OpenAI → BedrockPeak RSS high-water mark (MiB)1,198.7Valid—
OpenAI → BedrockTime to plateau (s)60Valid—
OpenAI → BedrockMemory growth (MiB/min)0.931Valid—
OpenAI → BedrockMemory load duration (s)300Valid—
OpenAI → BedrockMemory shape under load—Not measured—
OpenAI → BedrockIdle memory shape—Not measured—
OpenAI → BedrockIdle memory static1Valid—
OpenAI → BedrockIdle memory growth (MiB/min)1.08Valid—
OpenAI → BedrockThroughput (req/s)—Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound1 ms
OpenAI → BedrockThroughput (req/s)4,316concurrency 8; n=25,9005 ms
OpenAI → BedrockThroughput (req/s)4,440concurrency 16; n=26,64610 ms
OpenAI → BedrockThroughput (req/s)4,587concurrency 64; n=27,53050 ms
OpenAI → BedrockThroughput (req/s)4,587concurrency 64; n=27,530100 ms
OpenAI → BedrockThroughput (req/s)4,587concurrency 64; n=27,530Unbounded
Anthropic → Bedrock——Not declared: recorded note 1—
OpenAI → OpenAIAdded latency p50 (µs)718Valid · n=7,996 gateway / 262,593 direct—
OpenAI → OpenAIAdded latency p99 (µs)943Valid · n=7,996 gateway / 262,593 direct—
OpenAI → OpenAIGateway p99 at concurrency 1 (µs)971Valid · n=7,996—
OpenAI → OpenAIDirect-to-mock p99 at concurrency 1 (µs)28Valid · n=262,593—
OpenAI → OpenAICPU per request (µs)709.3Valid—
OpenAI → OpenAIRequests per CPU-second1,409.8Valid—
OpenAI → OpenAICost window concurrency8Valid—
OpenAI → OpenAICost window core utilisation0.733Valid—
OpenAI → OpenAICost window successful requests24,883Valid—
OpenAI → OpenAICost window throughput (req/s)4,146Valid—
OpenAI → OpenAICost window threads142Valid—
OpenAI → OpenAINon-voluntary context switches per request0.008Valid—
OpenAI → OpenAICost window major page faults0Valid—
OpenAI → OpenAIAdded time to first event p50 (µs)881Valid · n=100 gateway / 100 direct—
OpenAI → OpenAIAdded time to first event p99 (µs)1,103Valid · n=100 gateway / 100 direct—
OpenAI → OpenAIAdded inter-frame gap p50 (µs)2Valid · n=63 gateway / 63 direct—
OpenAI → OpenAIAdded inter-frame gap p99 (µs)—Below resolution · n=63 gateway / 63 direct—
OpenAI → OpenAIStreams sustained (frames/s)52,911.8Valid—
OpenAI → OpenAIStreams sustained1,688Valid—
OpenAI → OpenAIIdle RSS (MiB)1,163.5Valid—
OpenAI → OpenAISteady-state RSS (MiB)1,170.9Valid—
OpenAI → OpenAIRecovered RSS (MiB)1,172Valid—
OpenAI → OpenAIPeak RSS (MiB)1,181.8Valid—
OpenAI → OpenAIPeak RSS high-water mark (MiB)1,287.1Valid—
OpenAI → OpenAITime to plateau (s)66Valid—
OpenAI → OpenAIMemory growth (MiB/min)0.896Valid—
OpenAI → OpenAIMemory load duration (s)300Valid—
OpenAI → OpenAIMemory shape under load—Not measured—
OpenAI → OpenAIIdle memory shape—Not measured—
OpenAI → OpenAIIdle memory static1Valid—
OpenAI → OpenAIIdle memory growth (MiB/min)0.606Valid—
OpenAI → OpenAIThroughput (req/s)2,688concurrency 2; n=16,1311 ms
OpenAI → OpenAIThroughput (req/s)5,685concurrency 8; n=34,1205 ms
OpenAI → OpenAIThroughput (req/s)6,095concurrency 32; n=36,57610 ms
OpenAI → OpenAIThroughput (req/s)6,138concurrency 64; n=36,83250 ms
OpenAI → OpenAIThroughput (req/s)6,138concurrency 64; n=36,832100 ms
OpenAI → OpenAIThroughput (req/s)6,138concurrency 64; n=36,832Unbounded
Gemini → OpenAI——Not declared: recorded note 1—
Bedrock → OpenAI——Not declared: recorded note 1—
Cohere → OpenAI——Not declared: recorded note 1—
Anthropic → OpenAIAdded latency p50 (µs)954Valid · n=6,058 gateway / 259,722 direct—
Anthropic → OpenAIAdded latency p99 (µs)1,252Valid · n=6,058 gateway / 259,722 direct—
Anthropic → OpenAIGateway p99 at concurrency 1 (µs)1,280Valid · n=6,058—
Anthropic → OpenAIDirect-to-mock p99 at concurrency 1 (µs)28Valid · n=259,722—
Anthropic → OpenAICPU per request (µs)968.3Valid—
Anthropic → OpenAIRequests per CPU-second1,032.7Valid—
Anthropic → OpenAICost window concurrency8Valid—
Anthropic → OpenAICost window core utilisation0.501Valid—
Anthropic → OpenAICost window successful requests12,465Valid—
Anthropic → OpenAICost window throughput (req/s)2,076Valid—
Anthropic → OpenAICost window threads142Valid—
Anthropic → OpenAINon-voluntary context switches per request0.0000802Valid—
Anthropic → OpenAICost window major page faults0Valid—
Anthropic → OpenAIAdded time to first event p50 (µs)1,262Valid · n=100 gateway / 100 direct—
Anthropic → OpenAIAdded time to first event p99 (µs)1,705Valid · n=100 gateway / 100 direct—
Anthropic → OpenAIAdded inter-frame gap p50 (µs)8Valid · n=63 gateway / 63 direct—
Anthropic → OpenAIAdded inter-frame gap p99 (µs)163Valid · n=63 gateway / 63 direct—
Anthropic → OpenAIStreams sustained (frames/s)19,376.7Valid—
Anthropic → OpenAIStreams sustained5,088Valid—
Anthropic → OpenAIIdle RSS (MiB)1,151.7Valid—
Anthropic → OpenAISteady-state RSS (MiB)1,160.2Valid—
Anthropic → OpenAIRecovered RSS (MiB)1,161.2Valid—
Anthropic → OpenAIPeak RSS (MiB)1,177.2Valid—
Anthropic → OpenAIPeak RSS high-water mark (MiB)1,280.4Valid—
Anthropic → OpenAITime to plateau (s)60Valid—
Anthropic → OpenAIMemory growth (MiB/min)0.529Valid—
Anthropic → OpenAIMemory load duration (s)300Valid—
Anthropic → OpenAIMemory shape under load—Not measured—
Anthropic → OpenAIIdle memory shape—Not measured—
Anthropic → OpenAIIdle memory static1Valid—
Anthropic → OpenAIIdle memory growth (MiB/min)1Valid—
Anthropic → OpenAIThroughput (req/s)—Bound not met; concurrency not recorded: every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound1 ms
Anthropic → OpenAIThroughput (req/s)3,867concurrency 8; n=23,2115 ms
Anthropic → OpenAIThroughput (req/s)3,999concurrency 16; n=23,99810 ms
Anthropic → OpenAIThroughput (req/s)4,118concurrency 64; n=24,71350 ms
Anthropic → OpenAIThroughput (req/s)4,118concurrency 64; n=24,713100 ms
Anthropic → OpenAIThroughput (req/s)4,118concurrency 64; n=24,713Unbounded
OpenAI Responses → OpenAI——Not declared: recorded note 1—

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 → upstreamCell stateProbe HTTP statusRecorded reasonRecorded detailEvidence scopeResponse excerptTested build
Cohere → CohereServed———Recorded run evidence—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
Bedrock → CohereNot declared—recorded note 1recorded note 1recorded note 2—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
Gemini → CohereNot declared—recorded note 1recorded note 1recorded note 2—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
OpenAI → CohereNot served200failedprobed and answered HTTP 200, but the body is not a usable openai response (see body_snippet)Recorded run evidencethe 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
Anthropic → CohereNot served200failedprobed and answered HTTP 200, but the body is not a usable anthropic response (see body_snippet)Recorded run evidencethe 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
OpenAI Responses → CohereNot declared—recorded note 1recorded note 1recorded note 2—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
Cohere → AnthropicNot declared—recorded note 1recorded note 1recorded note 2—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
Gemini → AnthropicNot declared—recorded note 1recorded note 1recorded note 2—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
OpenAI → AnthropicServed———Recorded run evidence—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
Bedrock → AnthropicNot declared—recorded note 1recorded note 1recorded note 2—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
OpenAI Responses → AnthropicNot declared—recorded note 1recorded note 1recorded note 2—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
Anthropic → AnthropicServed———Recorded run evidence—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
OpenAI → OpenAI ResponsesNot declared—recorded note 1recorded note 1recorded note 2—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
Gemini → OpenAI ResponsesNot declared—recorded note 1recorded note 1recorded note 2—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
Cohere → OpenAI ResponsesNot declared—recorded note 1recorded note 1recorded note 2—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
Bedrock → OpenAI ResponsesNot declared—recorded note 1recorded note 1recorded note 2—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
OpenAI Responses → OpenAI ResponsesServed———Recorded run evidence—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
Anthropic → OpenAI ResponsesNot declared—recorded note 1recorded note 1recorded note 2—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
OpenAI → GeminiServed———Recorded run evidence—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
Gemini → GeminiServed———Recorded run evidence—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
Bedrock → GeminiNot declared—recorded note 1recorded note 1recorded note 2—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
Cohere → GeminiNot declared—recorded note 1recorded note 1recorded note 2—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
OpenAI Responses → GeminiNot declared—recorded note 1recorded note 1recorded note 2—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
Anthropic → GeminiServed———Recorded run evidence—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
OpenAI Responses → BedrockNot declared—recorded note 1recorded note 1recorded note 2—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
Gemini → BedrockNot declared—recorded note 1recorded note 1recorded note 2—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
Cohere → BedrockNot declared—recorded note 1recorded note 1recorded note 2—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
Bedrock → BedrockServed———Recorded run evidence—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
OpenAI → BedrockServed———Recorded run evidence—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
Anthropic → BedrockNot declared—recorded note 1recorded note 1recorded note 2—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
OpenAI → OpenAIServed———Recorded run evidence—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
Gemini → OpenAINot declared—recorded note 1recorded note 1recorded note 2—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
Bedrock → OpenAINot declared—recorded note 1recorded note 1recorded note 2—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
Cohere → OpenAINot declared—recorded note 1recorded note 1recorded note 2—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
Anthropic → OpenAIServed———Recorded run evidence—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
OpenAI Responses → OpenAINot declared—recorded note 1recorded note 1recorded note 2—higress-registry.cn-hangzhou.cr.aliyuncs.com/higress/all-in-one:2.2.5
Harness commit
d240badd966933fa19eec0606b2d1f3945f18ad2
Mock SHA-256
99d01f709d8d9c1f49055aa265d5eb7475264d6b513911fe317aee74a71190f8
Load generator SHA-256
not recorded
net.core.somaxconn
65,535
net.ipv4.tcp_max_syn_backlog
65,535

Stream sweep diagnostics

Per streaming-sweep concurrency, as recorded by the engine: lanes by final HTTP status, and lanes that received no response.

Ingress → upstreamConcurrencyLanes by HTTP statusLanes with no response
OpenAI → Anthropic1HTTP 200: 10
OpenAI → Anthropic2HTTP 200: 20
OpenAI → Anthropic4HTTP 200: 40
OpenAI → Anthropic8HTTP 200: 80
OpenAI → Anthropic16HTTP 200: 160
OpenAI → Anthropic32HTTP 200: 320
OpenAI → Anthropic64HTTP 200: 640
OpenAI → Anthropic128HTTP 200: 1280
OpenAI → Anthropic256HTTP 200: 2560
OpenAI → Anthropic512HTTP 200: 5120
OpenAI → Anthropic1,024HTTP 200: 1,0240
OpenAI → Anthropic2,048HTTP 200: 2,0480
OpenAI → Anthropic4,096HTTP 200: 4,0960
OpenAI → Anthropic8,192HTTP 200: 8,082110
OpenAI → Anthropic6,144HTTP 200: 6,1440
OpenAI → Anthropic7,168HTTP 200: 7,1680
OpenAI → Anthropic7,680HTTP 200: 7,6800
OpenAI → Anthropic7,424HTTP 200: 7,4240
OpenAI → Anthropic7,552HTTP 200: 7,5520
OpenAI → Anthropic7,488HTTP 200: 7,4880
OpenAI → Anthropic7,456HTTP 200: 7,4560
OpenAI → Anthropic7,440HTTP 200: 7,4400
OpenAI → Anthropic7,432HTTP 200: 7,4320
OpenAI → Anthropic7,436HTTP 200: 7,4360
OpenAI → Anthropic7,438HTTP 200: 7,4380
OpenAI → Anthropic7,439HTTP 200: 7,4390
OpenAI → Anthropic7,438HTTP 200: 7,4380
OpenAI → Anthropic7,438HTTP 200: 7,4380
OpenAI → Anthropic5,767HTTP 200: 5,7670
OpenAI → Anthropic5,767HTTP 200: 5,7670
OpenAI → Anthropic5,767HTTP 200: 5,7670
Anthropic → Anthropic1HTTP 200: 10
Anthropic → Anthropic2HTTP 200: 20
Anthropic → Anthropic4HTTP 200: 40
Anthropic → Anthropic8HTTP 200: 80
Anthropic → Anthropic16HTTP 200: 160
Anthropic → Anthropic32HTTP 200: 320
Anthropic → Anthropic64HTTP 200: 640
Anthropic → Anthropic128HTTP 200: 1280
Anthropic → Anthropic256HTTP 200: 2560
Anthropic → Anthropic512HTTP 200: 5120
Anthropic → Anthropic1,024HTTP 200: 1,0240
Anthropic → Anthropic2,048HTTP 200: 2,0480
Anthropic → Anthropic4,096HTTP 200: 4,0960
Anthropic → Anthropic8,192HTTP 200: 8,1920
Anthropic → Anthropic16,384HTTP 200: 16,3840
Anthropic → Anthropic32,768HTTP 200: 17,77114,997
Anthropic → Anthropic24,576HTTP 200: 17,5627,014
Anthropic → Anthropic20,480HTTP 200: 17,3053,175
Anthropic → Anthropic18,432HTTP 200: 17,810622
Anthropic → Anthropic17,408HTTP 200: 17,4080
Anthropic → Anthropic16,896HTTP 200: 16,8960
Anthropic → Anthropic16,640HTTP 200: 16,6400
Anthropic → Anthropic16,512HTTP 200: 16,5120
Anthropic → Anthropic16,448HTTP 200: 16,4480
Anthropic → Anthropic16,416HTTP 200: 16,4160
Anthropic → Anthropic16,400HTTP 200: 16,4000
Anthropic → Anthropic16,392HTTP 200: 16,3920
Anthropic → Anthropic16,388HTTP 200: 16,3880
Anthropic → Anthropic16,390HTTP 200: 16,3900
Anthropic → Anthropic16,391HTTP 200: 16,3910
Anthropic → Anthropic16,391HTTP 200: 16,3910
Anthropic → Anthropic16,391HTTP 200: 16,3910
OpenAI → OpenAI1HTTP 200: 10
OpenAI → OpenAI2HTTP 200: 20
OpenAI → OpenAI4HTTP 200: 40
OpenAI → OpenAI8HTTP 200: 80
OpenAI → OpenAI16HTTP 200: 160
OpenAI → OpenAI32HTTP 200: 320
OpenAI → OpenAI64HTTP 200: 640
OpenAI → OpenAI128HTTP 200: 1280
OpenAI → OpenAI256HTTP 200: 2560
OpenAI → OpenAI512HTTP 200: 5120
OpenAI → OpenAI1,024HTTP 200: 1,0240
OpenAI → OpenAI2,048HTTP 200: 2,0480
OpenAI → OpenAI1,536HTTP 200: 1,5360
OpenAI → OpenAI1,792HTTP 200: 1,7920
OpenAI → OpenAI1,664HTTP 200: 1,6640
OpenAI → OpenAI1,728HTTP 200: 1,7280
OpenAI → OpenAI1,696HTTP 200: 1,6960
OpenAI → OpenAI1,680HTTP 200: 1,6800
OpenAI → OpenAI1,688HTTP 200: 1,6880
OpenAI → OpenAI1,692HTTP 200: 1,6920
OpenAI → OpenAI1,690HTTP 200: 1,6900
OpenAI → OpenAI1,689HTTP 200: 1,6890
OpenAI → OpenAI1,688HTTP 200: 1,6880
OpenAI → OpenAI1,688HTTP 200: 1,6880
Anthropic → OpenAI1HTTP 200: 10
Anthropic → OpenAI2HTTP 200: 20
Anthropic → OpenAI4HTTP 200: 40
Anthropic → OpenAI8HTTP 200: 80
Anthropic → OpenAI16HTTP 200: 160
Anthropic → OpenAI32HTTP 200: 320
Anthropic → OpenAI64HTTP 200: 640
Anthropic → OpenAI128HTTP 200: 1280
Anthropic → OpenAI256HTTP 200: 2560
Anthropic → OpenAI512HTTP 200: 5120
Anthropic → OpenAI1,024HTTP 200: 1,0240
Anthropic → OpenAI2,048HTTP 200: 2,0480
Anthropic → OpenAI4,096HTTP 200: 4,0960
Anthropic → OpenAI8,192HTTP 200: 6,6881,504
Anthropic → OpenAI6,144HTTP 200: 6,1440
Anthropic → OpenAI5,120HTTP 200: 5,1200
Anthropic → OpenAI5,632HTTP 200: 5,6320
Anthropic → OpenAI5,888HTTP 200: 5,8880
Anthropic → OpenAI6,016HTTP 200: 6,0160
Anthropic → OpenAI6,080HTTP 200: 6,0800
Anthropic → OpenAI6,112HTTP 200: 6,1120
Anthropic → OpenAI6,096HTTP 200: 6,0960
Anthropic → OpenAI6,088HTTP 200: 6,0880
Anthropic → OpenAI6,084HTTP 200: 6,0840
Anthropic → OpenAI6,082HTTP 200: 6,0820
Anthropic → OpenAI6,081HTTP 200: 6,0810
Anthropic → OpenAI6,080HTTP 200: 6,0800
Anthropic → OpenAI6,080HTTP 200: 6,0800
Anthropic → OpenAI5,088HTTP 200: 5,0880
Anthropic → OpenAI5,088HTTP 200: 5,0880
Anthropic → OpenAI5,088HTTP 200: 5,0880

Recorded notes

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

  1. 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
  2. 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

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

Historical measurements