# agentgateway 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) .

Rust · recorded build: ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b. Measured: 2026-11-10T17:39:19Z.

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

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

## Published measurements

| Ingress → upstream | Metric | Value | State / reason | p99 bound (ms) |
| --- | --- | --- | --- | --- |
| openai-responses → openai-responses | added_latency_p50_us | 210 | valid; gateway_c1_samples=25353; direct_c1_samples=241872 | — |
| openai-responses → openai-responses | added_latency_p99_us | 247 | valid; gateway_c1_samples=25353; direct_c1_samples=241872 | — |
| openai-responses → openai-responses | gateway_c1_p99_us | 277 | valid; gateway_c1_samples=25353 | — |
| openai-responses → openai-responses | direct_c1_p99_us | 30 | valid; direct_c1_samples=241872 | — |
| openai-responses → openai-responses | cpu_us_per_request | 195.8575912414804 | valid | — |
| openai-responses → openai-responses | rps_per_cpu_second | 5105.750528541226 | valid | — |
| openai-responses → openai-responses | cost_window_conc | 8 | valid | — |
| openai-responses → openai-responses | cost_core_utilisation | 0.9823738537057497 | valid | — |
| openai-responses → openai-responses | cost_window_ok | 120751 | valid | — |
| openai-responses → openai-responses | cost_window_rps | 20122 | valid | — |
| openai-responses → openai-responses | cost_threads | 13 | valid | — |
| openai-responses → openai-responses | cost_nonvol_ctxt_per_request | 0 | 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 | 34.4609375 | valid | — |
| openai-responses → openai-responses | steady_state_rss_mib | 53.240234375 | valid | — |
| openai-responses → openai-responses | recovered_rss_mib | 50.75390625 | valid | — |
| openai-responses → openai-responses | peak_rss_mib | 54.40625 | valid | — |
| openai-responses → openai-responses | peak_rss_hwm_mib | 54.2421875 | valid | — |
| openai-responses → openai-responses | time_to_plateau_s | not_measured | not_measured | — |
| openai-responses → openai-responses | growth_rate_mib_per_min | 0.763626123026112 | valid | — |
| openai-responses → openai-responses | load_s | 300 | valid | — |
| openai-responses → openai-responses | shape | 0 | valid | — |
| openai-responses → openai-responses | idle_shape | not_measured | not_measured | — |
| openai-responses → openai-responses | idle_static | 1 | valid | — |
| openai-responses → openai-responses | idle_growth_rate_mib_per_min | 0.031671412126784065 | valid | — |
| openai-responses → openai-responses | throughput_rps | 19964 req/s | concurrency 8; samples 119803 | 1 |
| openai-responses → openai-responses | throughput_rps | 20925 req/s | concurrency 64; samples 125554 | 5 |
| openai-responses → openai-responses | throughput_rps | 21051 req/s | concurrency 128; samples 126327 | 10 |
| openai-responses → openai-responses | throughput_rps | 21224 req/s | concurrency 256; samples 127359 | 50 |
| openai-responses → openai-responses | throughput_rps | 21224 req/s | concurrency 256; samples 127359 | 100 |
| openai-responses → openai-responses | throughput_rps | 21224 req/s | concurrency 256; samples 127359 | unbounded |
| cohere → openai-responses | — | — | not_declared: Pinned v1.7.0 image and source 26f5aebe99a36a037fe54f8080a608aca5547e15: CHAT_TRANSLATIONS in crates/agentgateway/src/llm/mod.rs defines supported conversions. Separate custom providers constrain Chat versus Responses wire. Gemini native and Chat-to-Gemini are supported; Messages/Responses use Google OpenAI compatibility instead of native Gemini and are not counted as Gemini wire. Cohere preset similarly uses OpenAI compatibility, not Cohere native chat. No native Cohere or Bedrock ingress conversion is registered. Local recorder and response checks establish 12 pairs. Responses mock token-details correction is required for Messages-to-Responses. Evidence: results/diagnostics/refresh-configuration/agentgateway-protocol-preflight.json | — |
| openai → openai-responses | — | — | not_declared: Pinned v1.7.0 image and source 26f5aebe99a36a037fe54f8080a608aca5547e15: CHAT_TRANSLATIONS in crates/agentgateway/src/llm/mod.rs defines supported conversions. Separate custom providers constrain Chat versus Responses wire. Gemini native and Chat-to-Gemini are supported; Messages/Responses use Google OpenAI compatibility instead of native Gemini and are not counted as Gemini wire. Cohere preset similarly uses OpenAI compatibility, not Cohere native chat. No native Cohere or Bedrock ingress conversion is registered. Local recorder and response checks establish 12 pairs. Responses mock token-details correction is required for Messages-to-Responses. Evidence: results/diagnostics/refresh-configuration/agentgateway-protocol-preflight.json | — |
| anthropic → openai-responses | added_latency_p50_us | 227 | valid; gateway_c1_samples=23610; direct_c1_samples=229535 | — |
| anthropic → openai-responses | added_latency_p99_us | 267 | valid; gateway_c1_samples=23610; direct_c1_samples=229535 | — |
| anthropic → openai-responses | gateway_c1_p99_us | 299 | valid; gateway_c1_samples=23610 | — |
| anthropic → openai-responses | direct_c1_p99_us | 32 | valid; direct_c1_samples=229535 | — |
| anthropic → openai-responses | cpu_us_per_request | 207.76145582109186 | valid | — |
| anthropic → openai-responses | rps_per_cpu_second | 4813.212325875897 | valid | — |
| anthropic → openai-responses | cost_window_conc | 8 | valid | — |
| anthropic → openai-responses | cost_core_utilisation | 0.9840677551204744 | valid | — |
| anthropic → openai-responses | cost_window_ok | 114025 | valid | — |
| anthropic → openai-responses | cost_window_rps | 19002 | valid | — |
| anthropic → openai-responses | cost_threads | 13 | valid | — |
| anthropic → openai-responses | cost_nonvol_ctxt_per_request | 0 | valid | — |
| anthropic → openai-responses | cost_majflt | 0 | valid | — |
| anthropic → openai-responses | added_ttft_p50_us | untestable | untestable | — |
| anthropic → openai-responses | added_ttft_p99_us | untestable | untestable | — |
| anthropic → openai-responses | added_gap_p50_us | untestable | untestable | — |
| anthropic → openai-responses | added_gap_p99_us | untestable | untestable | — |
| anthropic → openai-responses | streams_sustained_fps | untestable | untestable | — |
| anthropic → openai-responses | streams_sustained | untestable | untestable | — |
| anthropic → openai-responses | idle_rss_mib | 34.33984375 | valid | — |
| anthropic → openai-responses | steady_state_rss_mib | 54.75 | valid | — |
| anthropic → openai-responses | recovered_rss_mib | 51.7734375 | valid | — |
| anthropic → openai-responses | peak_rss_mib | 55.41796875 | valid | — |
| anthropic → openai-responses | peak_rss_hwm_mib | 55.16015625 | valid | — |
| anthropic → openai-responses | time_to_plateau_s | 252 | valid | — |
| anthropic → openai-responses | growth_rate_mib_per_min | 0.175601960927026 | valid | — |
| anthropic → openai-responses | load_s | 300 | valid | — |
| anthropic → openai-responses | shape | not_measured | not_measured | — |
| anthropic → openai-responses | idle_shape | not_measured | not_measured | — |
| anthropic → openai-responses | idle_static | 1 | valid | — |
| anthropic → openai-responses | idle_growth_rate_mib_per_min | 0.04876842506033785 | valid | — |
| anthropic → openai-responses | throughput_rps | 18964 req/s | concurrency 8; samples 113798 | 1 |
| anthropic → openai-responses | throughput_rps | 19692 req/s | concurrency 32; samples 118168 | 5 |
| anthropic → openai-responses | throughput_rps | 19772 req/s | concurrency 64; samples 118635 | 10 |
| anthropic → openai-responses | throughput_rps | 20024 req/s | concurrency 256; samples 120161 | 50 |
| anthropic → openai-responses | throughput_rps | 20024 req/s | concurrency 256; samples 120161 | 100 |
| anthropic → openai-responses | throughput_rps | 20024 req/s | concurrency 256; samples 120161 | unbounded |
| bedrock → openai-responses | — | — | not_declared: Pinned v1.7.0 image and source 26f5aebe99a36a037fe54f8080a608aca5547e15: CHAT_TRANSLATIONS in crates/agentgateway/src/llm/mod.rs defines supported conversions. Separate custom providers constrain Chat versus Responses wire. Gemini native and Chat-to-Gemini are supported; Messages/Responses use Google OpenAI compatibility instead of native Gemini and are not counted as Gemini wire. Cohere preset similarly uses OpenAI compatibility, not Cohere native chat. No native Cohere or Bedrock ingress conversion is registered. Local recorder and response checks establish 12 pairs. Responses mock token-details correction is required for Messages-to-Responses. Evidence: results/diagnostics/refresh-configuration/agentgateway-protocol-preflight.json | — |
| gemini → openai-responses | — | — | not_declared: Pinned v1.7.0 image and source 26f5aebe99a36a037fe54f8080a608aca5547e15: CHAT_TRANSLATIONS in crates/agentgateway/src/llm/mod.rs defines supported conversions. Separate custom providers constrain Chat versus Responses wire. Gemini native and Chat-to-Gemini are supported; Messages/Responses use Google OpenAI compatibility instead of native Gemini and are not counted as Gemini wire. Cohere preset similarly uses OpenAI compatibility, not Cohere native chat. No native Cohere or Bedrock ingress conversion is registered. Local recorder and response checks establish 12 pairs. Responses mock token-details correction is required for Messages-to-Responses. Evidence: results/diagnostics/refresh-configuration/agentgateway-protocol-preflight.json | — |
| anthropic → bedrock | added_latency_p50_us | 218 | valid; gateway_c1_samples=24390; direct_c1_samples=237655 | — |
| anthropic → bedrock | added_latency_p99_us | 256 | valid; gateway_c1_samples=24390; direct_c1_samples=237655 | — |
| anthropic → bedrock | gateway_c1_p99_us | 287 | valid; gateway_c1_samples=24390 | — |
| anthropic → bedrock | direct_c1_p99_us | 31 | valid; direct_c1_samples=237655 | — |
| anthropic → bedrock | cpu_us_per_request | 200.20825044232055 | valid | — |
| anthropic → bedrock | rps_per_cpu_second | 4994.799154334039 | valid | — |
| anthropic → bedrock | cost_window_conc | 8 | valid | — |
| anthropic → bedrock | cost_core_utilisation | 0.9823076767856654 | valid | — |
| anthropic → bedrock | cost_window_ok | 118127 | valid | — |
| anthropic → bedrock | cost_window_rps | 19685 | valid | — |
| anthropic → bedrock | cost_threads | 13 | valid | — |
| anthropic → bedrock | cost_nonvol_ctxt_per_request | 0 | valid | — |
| anthropic → bedrock | cost_majflt | 0 | valid | — |
| anthropic → bedrock | added_ttft_p50_us | untestable | untestable | — |
| anthropic → bedrock | added_ttft_p99_us | untestable | untestable | — |
| anthropic → bedrock | added_gap_p50_us | untestable | untestable | — |
| anthropic → bedrock | added_gap_p99_us | untestable | untestable | — |
| anthropic → bedrock | streams_sustained_fps | untestable | untestable | — |
| anthropic → bedrock | streams_sustained | untestable | untestable | — |
| anthropic → bedrock | idle_rss_mib | 34.07421875 | valid | — |
| anthropic → bedrock | steady_state_rss_mib | 54.66015625 | valid | — |
| anthropic → bedrock | recovered_rss_mib | 51.2421875 | valid | — |
| anthropic → bedrock | peak_rss_mib | 55.44921875 | valid | — |
| anthropic → bedrock | peak_rss_hwm_mib | 55.2578125 | valid | — |
| anthropic → bedrock | time_to_plateau_s | not_measured | not_measured | — |
| anthropic → bedrock | growth_rate_mib_per_min | -1.302289476054103 | valid | — |
| anthropic → bedrock | load_s | 300 | valid | — |
| anthropic → bedrock | shape | -1 | valid | — |
| anthropic → bedrock | idle_shape | not_measured | not_measured | — |
| anthropic → bedrock | idle_static | 1 | valid | — |
| anthropic → bedrock | idle_growth_rate_mib_per_min | 0.04872452430819797 | valid | — |
| anthropic → bedrock | throughput_rps | 19560 req/s | concurrency 8; samples 117376 | 1 |
| anthropic → bedrock | throughput_rps | 20404 req/s | concurrency 64; samples 122433 | 5 |
| anthropic → bedrock | throughput_rps | 20404 req/s | concurrency 64; samples 122433 | 10 |
| anthropic → bedrock | throughput_rps | 20660 req/s | concurrency 256; samples 123971 | 50 |
| anthropic → bedrock | throughput_rps | 20660 req/s | concurrency 256; samples 123971 | 100 |
| anthropic → bedrock | throughput_rps | 20660 req/s | concurrency 256; samples 123971 | unbounded |
| gemini → bedrock | — | — | not_declared: Pinned v1.7.0 image and source 26f5aebe99a36a037fe54f8080a608aca5547e15: CHAT_TRANSLATIONS in crates/agentgateway/src/llm/mod.rs defines supported conversions. Separate custom providers constrain Chat versus Responses wire. Gemini native and Chat-to-Gemini are supported; Messages/Responses use Google OpenAI compatibility instead of native Gemini and are not counted as Gemini wire. Cohere preset similarly uses OpenAI compatibility, not Cohere native chat. No native Cohere or Bedrock ingress conversion is registered. Local recorder and response checks establish 12 pairs. Responses mock token-details correction is required for Messages-to-Responses. Evidence: results/diagnostics/refresh-configuration/agentgateway-protocol-preflight.json | — |
| cohere → bedrock | — | — | not_declared: Pinned v1.7.0 image and source 26f5aebe99a36a037fe54f8080a608aca5547e15: CHAT_TRANSLATIONS in crates/agentgateway/src/llm/mod.rs defines supported conversions. Separate custom providers constrain Chat versus Responses wire. Gemini native and Chat-to-Gemini are supported; Messages/Responses use Google OpenAI compatibility instead of native Gemini and are not counted as Gemini wire. Cohere preset similarly uses OpenAI compatibility, not Cohere native chat. No native Cohere or Bedrock ingress conversion is registered. Local recorder and response checks establish 12 pairs. Responses mock token-details correction is required for Messages-to-Responses. Evidence: results/diagnostics/refresh-configuration/agentgateway-protocol-preflight.json | — |
| openai-responses → bedrock | added_latency_p50_us | 231 | valid; gateway_c1_samples=23233; direct_c1_samples=241925 | — |
| openai-responses → bedrock | added_latency_p99_us | 270 | valid; gateway_c1_samples=23233; direct_c1_samples=241925 | — |
| openai-responses → bedrock | gateway_c1_p99_us | 301 | valid; gateway_c1_samples=23233 | — |
| openai-responses → bedrock | direct_c1_p99_us | 31 | valid; direct_c1_samples=241925 | — |
| openai-responses → bedrock | cpu_us_per_request | 208.7132725430598 | valid | — |
| openai-responses → bedrock | rps_per_cpu_second | 4791.262135922329 | valid | — |
| openai-responses → bedrock | cost_window_conc | 8 | valid | — |
| openai-responses → bedrock | cost_core_utilisation | 0.9840643588611752 | valid | — |
| openai-responses → bedrock | cost_window_ok | 113505 | valid | — |
| openai-responses → bedrock | cost_window_rps | 18915 | valid | — |
| openai-responses → bedrock | cost_threads | 13 | valid | — |
| openai-responses → bedrock | cost_nonvol_ctxt_per_request | 0 | valid | — |
| openai-responses → bedrock | cost_majflt | 0 | valid | — |
| openai-responses → bedrock | added_ttft_p50_us | untestable | untestable | — |
| openai-responses → bedrock | added_ttft_p99_us | untestable | untestable | — |
| openai-responses → bedrock | added_gap_p50_us | untestable | untestable | — |
| openai-responses → bedrock | added_gap_p99_us | untestable | untestable | — |
| openai-responses → bedrock | streams_sustained_fps | untestable | untestable | — |
| openai-responses → bedrock | streams_sustained | untestable | untestable | — |
| openai-responses → bedrock | idle_rss_mib | 34.09765625 | valid | — |
| openai-responses → bedrock | steady_state_rss_mib | 54.751953125 | valid | — |
| openai-responses → bedrock | recovered_rss_mib | 51.9921875 | valid | — |
| openai-responses → bedrock | peak_rss_mib | 55.734375 | valid | — |
| openai-responses → bedrock | peak_rss_hwm_mib | 55.29296875 | valid | — |
| openai-responses → bedrock | time_to_plateau_s | not_measured | not_measured | — |
| openai-responses → bedrock | growth_rate_mib_per_min | 0.02300161895283871 | valid | — |
| openai-responses → bedrock | load_s | 300 | valid | — |
| openai-responses → bedrock | shape | 0 | valid | — |
| openai-responses → bedrock | idle_shape | not_measured | not_measured | — |
| openai-responses → bedrock | idle_static | 1 | valid | — |
| openai-responses → bedrock | idle_growth_rate_mib_per_min | 0.014420287700335939 | valid | — |
| openai-responses → bedrock | throughput_rps | 18692 req/s | concurrency 8; samples 112168 | 1 |
| openai-responses → bedrock | throughput_rps | 19412 req/s | concurrency 32; samples 116482 | 5 |
| openai-responses → bedrock | throughput_rps | 19442 req/s | concurrency 64; samples 116658 | 10 |
| openai-responses → bedrock | throughput_rps | 19736 req/s | concurrency 256; samples 118430 | 50 |
| openai-responses → bedrock | throughput_rps | 19736 req/s | concurrency 256; samples 118430 | 100 |
| openai-responses → bedrock | throughput_rps | 19736 req/s | concurrency 256; samples 118430 | unbounded |
| bedrock → bedrock | — | — | not_declared: Pinned v1.7.0 image and source 26f5aebe99a36a037fe54f8080a608aca5547e15: CHAT_TRANSLATIONS in crates/agentgateway/src/llm/mod.rs defines supported conversions. Separate custom providers constrain Chat versus Responses wire. Gemini native and Chat-to-Gemini are supported; Messages/Responses use Google OpenAI compatibility instead of native Gemini and are not counted as Gemini wire. Cohere preset similarly uses OpenAI compatibility, not Cohere native chat. No native Cohere or Bedrock ingress conversion is registered. Local recorder and response checks establish 12 pairs. Responses mock token-details correction is required for Messages-to-Responses. Evidence: results/diagnostics/refresh-configuration/agentgateway-protocol-preflight.json | — |
| openai → bedrock | added_latency_p50_us | 219 | valid; gateway_c1_samples=24338; direct_c1_samples=241582 | — |
| openai → bedrock | added_latency_p99_us | 256 | valid; gateway_c1_samples=24338; direct_c1_samples=241582 | — |
| openai → bedrock | gateway_c1_p99_us | 286 | valid; gateway_c1_samples=24338 | — |
| openai → bedrock | direct_c1_p99_us | 30 | valid; direct_c1_samples=241582 | — |
| openai → bedrock | cpu_us_per_request | 205.61558824805797 | valid | — |
| openai → bedrock | rps_per_cpu_second | 4863.4444913465595 | valid | — |
| openai → bedrock | cost_window_conc | 8 | valid | — |
| openai → bedrock | cost_core_utilisation | 0.9840817944727618 | valid | — |
| openai → bedrock | cost_window_ok | 115215 | valid | — |
| openai → bedrock | cost_window_rps | 19200 | valid | — |
| openai → bedrock | cost_threads | 13 | valid | — |
| openai → bedrock | cost_nonvol_ctxt_per_request | 0 | 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 | 34.1953125 | valid | — |
| openai → bedrock | steady_state_rss_mib | 53.9375 | valid | — |
| openai → bedrock | recovered_rss_mib | 50.9140625 | valid | — |
| openai → bedrock | peak_rss_mib | 54.71875 | valid | — |
| openai → bedrock | peak_rss_hwm_mib | 54.3125 | valid | — |
| openai → bedrock | time_to_plateau_s | not_measured | not_measured | — |
| openai → bedrock | growth_rate_mib_per_min | -0.6369156722067312 | valid | — |
| openai → bedrock | load_s | 300 | valid | — |
| openai → bedrock | shape | -1 | valid | — |
| openai → bedrock | idle_shape | not_measured | not_measured | — |
| openai → bedrock | idle_static | 1 | valid | — |
| openai → bedrock | idle_growth_rate_mib_per_min | 0.0445382197398453 | valid | — |
| openai → bedrock | throughput_rps | 19159 req/s | concurrency 8; samples 114971 | 1 |
| openai → bedrock | throughput_rps | 19912 req/s | concurrency 32; samples 119487 | 5 |
| openai → bedrock | throughput_rps | 20007 req/s | concurrency 64; samples 120063 | 10 |
| openai → bedrock | throughput_rps | 20265 req/s | concurrency 256; samples 121609 | 50 |
| openai → bedrock | throughput_rps | 20265 req/s | concurrency 256; samples 121609 | 100 |
| openai → bedrock | throughput_rps | 20265 req/s | concurrency 256; samples 121609 | unbounded |
| anthropic → cohere | — | — | not_declared: Pinned v1.7.0 image and source 26f5aebe99a36a037fe54f8080a608aca5547e15: CHAT_TRANSLATIONS in crates/agentgateway/src/llm/mod.rs defines supported conversions. Separate custom providers constrain Chat versus Responses wire. Gemini native and Chat-to-Gemini are supported; Messages/Responses use Google OpenAI compatibility instead of native Gemini and are not counted as Gemini wire. Cohere preset similarly uses OpenAI compatibility, not Cohere native chat. No native Cohere or Bedrock ingress conversion is registered. Local recorder and response checks establish 12 pairs. Responses mock token-details correction is required for Messages-to-Responses. Evidence: results/diagnostics/refresh-configuration/agentgateway-protocol-preflight.json | — |
| openai-responses → cohere | — | — | not_declared: Pinned v1.7.0 image and source 26f5aebe99a36a037fe54f8080a608aca5547e15: CHAT_TRANSLATIONS in crates/agentgateway/src/llm/mod.rs defines supported conversions. Separate custom providers constrain Chat versus Responses wire. Gemini native and Chat-to-Gemini are supported; Messages/Responses use Google OpenAI compatibility instead of native Gemini and are not counted as Gemini wire. Cohere preset similarly uses OpenAI compatibility, not Cohere native chat. No native Cohere or Bedrock ingress conversion is registered. Local recorder and response checks establish 12 pairs. Responses mock token-details correction is required for Messages-to-Responses. Evidence: results/diagnostics/refresh-configuration/agentgateway-protocol-preflight.json | — |
| gemini → cohere | — | — | not_declared: Pinned v1.7.0 image and source 26f5aebe99a36a037fe54f8080a608aca5547e15: CHAT_TRANSLATIONS in crates/agentgateway/src/llm/mod.rs defines supported conversions. Separate custom providers constrain Chat versus Responses wire. Gemini native and Chat-to-Gemini are supported; Messages/Responses use Google OpenAI compatibility instead of native Gemini and are not counted as Gemini wire. Cohere preset similarly uses OpenAI compatibility, not Cohere native chat. No native Cohere or Bedrock ingress conversion is registered. Local recorder and response checks establish 12 pairs. Responses mock token-details correction is required for Messages-to-Responses. Evidence: results/diagnostics/refresh-configuration/agentgateway-protocol-preflight.json | — |
| cohere → cohere | — | — | not_declared: Pinned v1.7.0 image and source 26f5aebe99a36a037fe54f8080a608aca5547e15: CHAT_TRANSLATIONS in crates/agentgateway/src/llm/mod.rs defines supported conversions. Separate custom providers constrain Chat versus Responses wire. Gemini native and Chat-to-Gemini are supported; Messages/Responses use Google OpenAI compatibility instead of native Gemini and are not counted as Gemini wire. Cohere preset similarly uses OpenAI compatibility, not Cohere native chat. No native Cohere or Bedrock ingress conversion is registered. Local recorder and response checks establish 12 pairs. Responses mock token-details correction is required for Messages-to-Responses. Evidence: results/diagnostics/refresh-configuration/agentgateway-protocol-preflight.json | — |
| openai → cohere | — | — | not_declared: Pinned v1.7.0 image and source 26f5aebe99a36a037fe54f8080a608aca5547e15: CHAT_TRANSLATIONS in crates/agentgateway/src/llm/mod.rs defines supported conversions. Separate custom providers constrain Chat versus Responses wire. Gemini native and Chat-to-Gemini are supported; Messages/Responses use Google OpenAI compatibility instead of native Gemini and are not counted as Gemini wire. Cohere preset similarly uses OpenAI compatibility, not Cohere native chat. No native Cohere or Bedrock ingress conversion is registered. Local recorder and response checks establish 12 pairs. Responses mock token-details correction is required for Messages-to-Responses. Evidence: results/diagnostics/refresh-configuration/agentgateway-protocol-preflight.json | — |
| bedrock → cohere | — | — | not_declared: Pinned v1.7.0 image and source 26f5aebe99a36a037fe54f8080a608aca5547e15: CHAT_TRANSLATIONS in crates/agentgateway/src/llm/mod.rs defines supported conversions. Separate custom providers constrain Chat versus Responses wire. Gemini native and Chat-to-Gemini are supported; Messages/Responses use Google OpenAI compatibility instead of native Gemini and are not counted as Gemini wire. Cohere preset similarly uses OpenAI compatibility, not Cohere native chat. No native Cohere or Bedrock ingress conversion is registered. Local recorder and response checks establish 12 pairs. Responses mock token-details correction is required for Messages-to-Responses. Evidence: results/diagnostics/refresh-configuration/agentgateway-protocol-preflight.json | — |
| openai-responses → openai | added_latency_p50_us | 229 | valid; gateway_c1_samples=23446; direct_c1_samples=240137 | — |
| openai-responses → openai | added_latency_p99_us | 267 | valid; gateway_c1_samples=23446; direct_c1_samples=240137 | — |
| openai-responses → openai | gateway_c1_p99_us | 298 | valid; gateway_c1_samples=23446 | — |
| openai-responses → openai | direct_c1_p99_us | 31 | valid; direct_c1_samples=240137 | — |
| openai-responses → openai | cpu_us_per_request | 207.54121914039914 | valid | — |
| openai-responses → openai | rps_per_cpu_second | 4818.319966230477 | valid | — |
| openai-responses → openai | cost_window_conc | 8 | valid | — |
| openai-responses → openai | cost_core_utilisation | 0.9840504611107183 | valid | — |
| openai-responses → openai | cost_window_ok | 114146 | valid | — |
| openai-responses → openai | cost_window_rps | 19022 | valid | — |
| openai-responses → openai | cost_threads | 13 | valid | — |
| openai-responses → openai | cost_nonvol_ctxt_per_request | 0 | valid | — |
| openai-responses → openai | cost_majflt | 0 | valid | — |
| openai-responses → openai | added_ttft_p50_us | untestable | untestable | — |
| openai-responses → openai | added_ttft_p99_us | untestable | untestable | — |
| openai-responses → openai | added_gap_p50_us | untestable | untestable | — |
| openai-responses → openai | added_gap_p99_us | untestable | untestable | — |
| openai-responses → openai | streams_sustained_fps | untestable | untestable | — |
| openai-responses → openai | streams_sustained | untestable | untestable | — |
| openai-responses → openai | idle_rss_mib | 34.03515625 | valid | — |
| openai-responses → openai | steady_state_rss_mib | 53.9375 | valid | — |
| openai-responses → openai | recovered_rss_mib | 51.01171875 | valid | — |
| openai-responses → openai | peak_rss_mib | 54.8046875 | valid | — |
| openai-responses → openai | peak_rss_hwm_mib | 54.5703125 | valid | — |
| openai-responses → openai | time_to_plateau_s | not_measured | not_measured | — |
| openai-responses → openai | growth_rate_mib_per_min | 1.1853562210621402 | valid | — |
| openai-responses → openai | load_s | 300 | valid | — |
| openai-responses → openai | shape | 1 | valid | — |
| openai-responses → openai | idle_shape | not_measured | not_measured | — |
| openai-responses → openai | idle_static | 1 | valid | — |
| openai-responses → openai | idle_growth_rate_mib_per_min | 0.045130039098198085 | valid | — |
| openai-responses → openai | throughput_rps | 19011 req/s | concurrency 8; samples 114080 | 1 |
| openai-responses → openai | throughput_rps | 19731 req/s | concurrency 32; samples 118399 | 5 |
| openai-responses → openai | throughput_rps | 19868 req/s | concurrency 64; samples 119220 | 10 |
| openai-responses → openai | throughput_rps | 20063 req/s | concurrency 256; samples 120396 | 50 |
| openai-responses → openai | throughput_rps | 20063 req/s | concurrency 256; samples 120396 | 100 |
| openai-responses → openai | throughput_rps | 20063 req/s | concurrency 256; samples 120396 | unbounded |
| cohere → openai | — | — | not_declared: Pinned v1.7.0 image and source 26f5aebe99a36a037fe54f8080a608aca5547e15: CHAT_TRANSLATIONS in crates/agentgateway/src/llm/mod.rs defines supported conversions. Separate custom providers constrain Chat versus Responses wire. Gemini native and Chat-to-Gemini are supported; Messages/Responses use Google OpenAI compatibility instead of native Gemini and are not counted as Gemini wire. Cohere preset similarly uses OpenAI compatibility, not Cohere native chat. No native Cohere or Bedrock ingress conversion is registered. Local recorder and response checks establish 12 pairs. Responses mock token-details correction is required for Messages-to-Responses. Evidence: results/diagnostics/refresh-configuration/agentgateway-protocol-preflight.json | — |
| openai → openai | added_latency_p50_us | 216 | valid; gateway_c1_samples=24618; direct_c1_samples=233879 | — |
| openai → openai | added_latency_p99_us | 251 | valid; gateway_c1_samples=24618; direct_c1_samples=233879 | — |
| openai → openai | gateway_c1_p99_us | 282 | valid; gateway_c1_samples=24618 | — |
| openai → openai | direct_c1_p99_us | 31 | valid; direct_c1_samples=233879 | — |
| openai → openai | cpu_us_per_request | 188.57584939720198 | valid | — |
| openai → openai | rps_per_cpu_second | 5302.905982905983 | valid | — |
| openai → openai | cost_window_conc | 8 | valid | — |
| openai → openai | cost_core_utilisation | 0.9718871854587116 | valid | — |
| openai → openai | cost_window_ok | 124088 | valid | — |
| openai → openai | cost_window_rps | 20679 | valid | — |
| openai → openai | cost_threads | 13 | valid | — |
| openai → openai | cost_nonvol_ctxt_per_request | 0 | valid | — |
| openai → openai | cost_majflt | 0 | valid | — |
| openai → openai | added_ttft_p50_us | 327 | valid; ttft_gw_samples=100; ttft_direct_samples=100 | — |
| openai → openai | added_ttft_p99_us | 360 | valid; ttft_gw_samples=100; ttft_direct_samples=100 | — |
| openai → openai | added_gap_p50_us | 10 | 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 | 32614.06218906021 | valid | — |
| openai → openai | streams_sustained | 2048 | valid | — |
| openai → openai | idle_rss_mib | 34.00390625 | valid | — |
| openai → openai | steady_state_rss_mib | 53.4921875 | valid | — |
| openai → openai | recovered_rss_mib | 49.99609375 | valid | — |
| openai → openai | peak_rss_mib | 54.5234375 | valid | — |
| openai → openai | peak_rss_hwm_mib | 54.2265625 | valid | — |
| openai → openai | time_to_plateau_s | not_measured | not_measured | — |
| openai → openai | growth_rate_mib_per_min | -2.0749110660555545 | valid | — |
| openai → openai | load_s | 300 | valid | — |
| openai → openai | shape | -1 | valid | — |
| openai → openai | idle_shape | not_measured | not_measured | — |
| openai → openai | idle_static | 1 | valid | — |
| openai → openai | idle_growth_rate_mib_per_min | 0.03612607648971414 | valid | — |
| openai → openai | throughput_rps | 21156 req/s | concurrency 8; samples 126952 | 1 |
| openai → openai | throughput_rps | 22239 req/s | concurrency 64; samples 133447 | 5 |
| openai → openai | throughput_rps | 22380 req/s | concurrency 128; samples 134304 | 10 |
| openai → openai | throughput_rps | 22545 req/s | concurrency 256; samples 135283 | 50 |
| openai → openai | throughput_rps | 22545 req/s | concurrency 256; samples 135283 | 100 |
| openai → openai | throughput_rps | 22545 req/s | concurrency 256; samples 135283 | unbounded |
| anthropic → openai | added_latency_p50_us | 212 | valid; gateway_c1_samples=25009; direct_c1_samples=233282 | — |
| anthropic → openai | added_latency_p99_us | 247 | valid; gateway_c1_samples=25009; direct_c1_samples=233282 | — |
| anthropic → openai | gateway_c1_p99_us | 279 | valid; gateway_c1_samples=25009 | — |
| anthropic → openai | direct_c1_p99_us | 32 | valid; direct_c1_samples=233282 | — |
| anthropic → openai | cpu_us_per_request | 196.34930127951347 | valid | — |
| anthropic → openai | rps_per_cpu_second | 5092.9643929643935 | valid | — |
| anthropic → openai | cost_window_conc | 8 | valid | — |
| anthropic → openai | cost_core_utilisation | 0.9681526005189147 | valid | — |
| anthropic → openai | cost_window_ok | 118717 | valid | — |
| anthropic → openai | cost_window_rps | 19784 | valid | — |
| anthropic → openai | cost_threads | 13 | valid | — |
| anthropic → openai | cost_nonvol_ctxt_per_request | 0 | valid | — |
| anthropic → openai | cost_majflt | 0 | valid | — |
| anthropic → openai | added_ttft_p50_us | 341 | valid; ttft_gw_samples=100; ttft_direct_samples=100 | — |
| anthropic → openai | added_ttft_p99_us | 395 | valid; ttft_gw_samples=100; ttft_direct_samples=100 | — |
| anthropic → openai | added_gap_p50_us | 3 | valid; gap_gw_samples=63; gap_direct_samples=63 | — |
| anthropic → openai | added_gap_p99_us | 17 | valid; gap_gw_samples=63; gap_direct_samples=63 | — |
| anthropic → openai | streams_sustained_fps | not_measured | not_measured | — |
| anthropic → openai | streams_sustained | not_measured | not_measured | — |
| anthropic → openai | idle_rss_mib | 33.6953125 | valid | — |
| anthropic → openai | steady_state_rss_mib | 53.57421875 | valid | — |
| anthropic → openai | recovered_rss_mib | 49.90234375 | valid | — |
| anthropic → openai | peak_rss_mib | 54.2109375 | valid | — |
| anthropic → openai | peak_rss_hwm_mib | 54.0546875 | valid | — |
| anthropic → openai | time_to_plateau_s | 156 | valid | — |
| anthropic → openai | growth_rate_mib_per_min | 0.10023099694236136 | 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 | 0.04929642550447049 | valid | — |
| anthropic → openai | throughput_rps | 19933 req/s | concurrency 8; samples 119616 | 1 |
| anthropic → openai | throughput_rps | 20841 req/s | concurrency 64; samples 125053 | 5 |
| anthropic → openai | throughput_rps | 20997 req/s | concurrency 128; samples 126001 | 10 |
| anthropic → openai | throughput_rps | 21100 req/s | concurrency 256; samples 126618 | 50 |
| anthropic → openai | throughput_rps | 21100 req/s | concurrency 256; samples 126618 | 100 |
| anthropic → openai | throughput_rps | 21100 req/s | concurrency 256; samples 126618 | unbounded |
| bedrock → openai | — | — | not_declared: Pinned v1.7.0 image and source 26f5aebe99a36a037fe54f8080a608aca5547e15: CHAT_TRANSLATIONS in crates/agentgateway/src/llm/mod.rs defines supported conversions. Separate custom providers constrain Chat versus Responses wire. Gemini native and Chat-to-Gemini are supported; Messages/Responses use Google OpenAI compatibility instead of native Gemini and are not counted as Gemini wire. Cohere preset similarly uses OpenAI compatibility, not Cohere native chat. No native Cohere or Bedrock ingress conversion is registered. Local recorder and response checks establish 12 pairs. Responses mock token-details correction is required for Messages-to-Responses. Evidence: results/diagnostics/refresh-configuration/agentgateway-protocol-preflight.json | — |
| gemini → openai | — | — | not_declared: Pinned v1.7.0 image and source 26f5aebe99a36a037fe54f8080a608aca5547e15: CHAT_TRANSLATIONS in crates/agentgateway/src/llm/mod.rs defines supported conversions. Separate custom providers constrain Chat versus Responses wire. Gemini native and Chat-to-Gemini are supported; Messages/Responses use Google OpenAI compatibility instead of native Gemini and are not counted as Gemini wire. Cohere preset similarly uses OpenAI compatibility, not Cohere native chat. No native Cohere or Bedrock ingress conversion is registered. Local recorder and response checks establish 12 pairs. Responses mock token-details correction is required for Messages-to-Responses. Evidence: results/diagnostics/refresh-configuration/agentgateway-protocol-preflight.json | — |
| anthropic → anthropic | added_latency_p50_us | 206 | valid; gateway_c1_samples=25747; direct_c1_samples=233859 | — |
| anthropic → anthropic | added_latency_p99_us | 240 | valid; gateway_c1_samples=25747; direct_c1_samples=233859 | — |
| anthropic → anthropic | gateway_c1_p99_us | 271 | valid; gateway_c1_samples=25747 | — |
| anthropic → anthropic | direct_c1_p99_us | 31 | valid; direct_c1_samples=233859 | — |
| anthropic → anthropic | cpu_us_per_request | 191.31926970324474 | valid | — |
| anthropic → anthropic | rps_per_cpu_second | 5226.865028029322 | valid | — |
| anthropic → anthropic | cost_window_conc | 8 | valid | — |
| anthropic → anthropic | cost_core_utilisation | 0.963164850868556 | valid | — |
| anthropic → anthropic | cost_window_ok | 121211 | valid | — |
| anthropic → anthropic | cost_window_rps | 20199 | valid | — |
| anthropic → anthropic | cost_threads | 13 | valid | — |
| anthropic → anthropic | cost_nonvol_ctxt_per_request | 0 | valid | — |
| anthropic → anthropic | cost_majflt | 0 | valid | — |
| anthropic → anthropic | added_ttft_p50_us | 323 | valid; ttft_gw_samples=100; ttft_direct_samples=100 | — |
| anthropic → anthropic | added_ttft_p99_us | 362 | valid; ttft_gw_samples=100; ttft_direct_samples=100 | — |
| anthropic → anthropic | added_gap_p50_us | 3 | valid; gap_gw_samples=63; gap_direct_samples=63 | — |
| anthropic → anthropic | added_gap_p99_us | below_resolution | below_resolution; gap_gw_samples=63; gap_direct_samples=63 | — |
| anthropic → anthropic | streams_sustained_fps | 29596.665864916016 | valid | — |
| anthropic → anthropic | streams_sustained | 4104 | valid | — |
| anthropic → anthropic | idle_rss_mib | 34.15625 | valid | — |
| anthropic → anthropic | steady_state_rss_mib | 53.16796875 | valid | — |
| anthropic → anthropic | recovered_rss_mib | 50.18359375 | valid | — |
| anthropic → anthropic | peak_rss_mib | 54.41015625 | valid | — |
| anthropic → anthropic | peak_rss_hwm_mib | 54.24609375 | valid | — |
| anthropic → anthropic | time_to_plateau_s | not_measured | not_measured | — |
| anthropic → anthropic | growth_rate_mib_per_min | -0.23727509558911336 | valid | — |
| anthropic → anthropic | load_s | 300 | valid | — |
| anthropic → anthropic | shape | 0 | valid | — |
| anthropic → anthropic | idle_shape | not_measured | not_measured | — |
| anthropic → anthropic | idle_static | 1 | valid | — |
| anthropic → anthropic | idle_growth_rate_mib_per_min | 0.048736586694404796 | valid | — |
| anthropic → anthropic | throughput_rps | 20502 req/s | concurrency 8; samples 123033 | 1 |
| anthropic → anthropic | throughput_rps | 21665 req/s | concurrency 64; samples 130000 | 5 |
| anthropic → anthropic | throughput_rps | 21795 req/s | concurrency 128; samples 130775 | 10 |
| anthropic → anthropic | throughput_rps | 21982 req/s | concurrency 256; samples 131913 | 50 |
| anthropic → anthropic | throughput_rps | 21982 req/s | concurrency 256; samples 131913 | 100 |
| anthropic → anthropic | throughput_rps | 21982 req/s | concurrency 256; samples 131913 | unbounded |
| bedrock → anthropic | — | — | not_declared: Pinned v1.7.0 image and source 26f5aebe99a36a037fe54f8080a608aca5547e15: CHAT_TRANSLATIONS in crates/agentgateway/src/llm/mod.rs defines supported conversions. Separate custom providers constrain Chat versus Responses wire. Gemini native and Chat-to-Gemini are supported; Messages/Responses use Google OpenAI compatibility instead of native Gemini and are not counted as Gemini wire. Cohere preset similarly uses OpenAI compatibility, not Cohere native chat. No native Cohere or Bedrock ingress conversion is registered. Local recorder and response checks establish 12 pairs. Responses mock token-details correction is required for Messages-to-Responses. Evidence: results/diagnostics/refresh-configuration/agentgateway-protocol-preflight.json | — |
| openai-responses → anthropic | — | — | not_declared: Pinned v1.7.0 image and source 26f5aebe99a36a037fe54f8080a608aca5547e15: CHAT_TRANSLATIONS in crates/agentgateway/src/llm/mod.rs defines supported conversions. Separate custom providers constrain Chat versus Responses wire. Gemini native and Chat-to-Gemini are supported; Messages/Responses use Google OpenAI compatibility instead of native Gemini and are not counted as Gemini wire. Cohere preset similarly uses OpenAI compatibility, not Cohere native chat. No native Cohere or Bedrock ingress conversion is registered. Local recorder and response checks establish 12 pairs. Responses mock token-details correction is required for Messages-to-Responses. Evidence: results/diagnostics/refresh-configuration/agentgateway-protocol-preflight.json | — |
| openai → anthropic | added_latency_p50_us | 228 | valid; gateway_c1_samples=23411; direct_c1_samples=237919 | — |
| openai → anthropic | added_latency_p99_us | 268 | valid; gateway_c1_samples=23411; direct_c1_samples=237919 | — |
| openai → anthropic | gateway_c1_p99_us | 299 | valid; gateway_c1_samples=23411 | — |
| openai → anthropic | direct_c1_p99_us | 31 | valid; direct_c1_samples=237919 | — |
| openai → anthropic | cpu_us_per_request | 206.99208794721747 | valid | — |
| openai → anthropic | rps_per_cpu_second | 4831.102531102531 | valid | — |
| openai → anthropic | cost_window_conc | 8 | valid | — |
| openai → anthropic | cost_core_utilisation | 0.9682216923673016 | valid | — |
| openai → anthropic | cost_window_ok | 112613 | valid | — |
| openai → anthropic | cost_window_rps | 18767 | valid | — |
| openai → anthropic | cost_threads | 13 | valid | — |
| openai → anthropic | cost_nonvol_ctxt_per_request | 0 | valid | — |
| openai → anthropic | cost_majflt | 0 | valid | — |
| openai → anthropic | added_ttft_p50_us | 405 | valid; ttft_gw_samples=100; ttft_direct_samples=100 | — |
| openai → anthropic | added_ttft_p99_us | 446 | valid; ttft_gw_samples=100; ttft_direct_samples=100 | — |
| openai → anthropic | added_gap_p50_us | 6 | valid; gap_gw_samples=63; gap_direct_samples=63 | — |
| openai → anthropic | added_gap_p99_us | 22 | valid; gap_gw_samples=63; gap_direct_samples=63 | — |
| openai → anthropic | streams_sustained_fps | 18079.394815032054 | valid | — |
| openai → anthropic | streams_sustained | 2559 | valid | — |
| openai → anthropic | idle_rss_mib | 34.1953125 | valid | — |
| openai → anthropic | steady_state_rss_mib | 53.81640625 | valid | — |
| openai → anthropic | recovered_rss_mib | 50.52734375 | valid | — |
| openai → anthropic | peak_rss_mib | 54.7109375 | valid | — |
| openai → anthropic | peak_rss_hwm_mib | 54.3515625 | valid | — |
| openai → anthropic | time_to_plateau_s | not_measured | not_measured | — |
| openai → anthropic | growth_rate_mib_per_min | -0.7642133230320394 | valid | — |
| openai → anthropic | load_s | 300 | valid | — |
| openai → anthropic | shape | 0 | valid | — |
| openai → anthropic | idle_shape | not_measured | not_measured | — |
| openai → anthropic | idle_static | 1 | valid | — |
| openai → anthropic | idle_growth_rate_mib_per_min | 0.04117718673808115 | valid | — |
| openai → anthropic | throughput_rps | 19127 req/s | concurrency 8; samples 114783 | 1 |
| openai → anthropic | throughput_rps | 19852 req/s | concurrency 32; samples 119122 | 5 |
| openai → anthropic | throughput_rps | 19973 req/s | concurrency 64; samples 119864 | 10 |
| openai → anthropic | throughput_rps | 20228 req/s | concurrency 256; samples 121382 | 50 |
| openai → anthropic | throughput_rps | 20228 req/s | concurrency 256; samples 121382 | 100 |
| openai → anthropic | throughput_rps | 20228 req/s | concurrency 256; samples 121382 | unbounded |
| cohere → anthropic | — | — | not_declared: Pinned v1.7.0 image and source 26f5aebe99a36a037fe54f8080a608aca5547e15: CHAT_TRANSLATIONS in crates/agentgateway/src/llm/mod.rs defines supported conversions. Separate custom providers constrain Chat versus Responses wire. Gemini native and Chat-to-Gemini are supported; Messages/Responses use Google OpenAI compatibility instead of native Gemini and are not counted as Gemini wire. Cohere preset similarly uses OpenAI compatibility, not Cohere native chat. No native Cohere or Bedrock ingress conversion is registered. Local recorder and response checks establish 12 pairs. Responses mock token-details correction is required for Messages-to-Responses. Evidence: results/diagnostics/refresh-configuration/agentgateway-protocol-preflight.json | — |
| gemini → anthropic | — | — | not_declared: Pinned v1.7.0 image and source 26f5aebe99a36a037fe54f8080a608aca5547e15: CHAT_TRANSLATIONS in crates/agentgateway/src/llm/mod.rs defines supported conversions. Separate custom providers constrain Chat versus Responses wire. Gemini native and Chat-to-Gemini are supported; Messages/Responses use Google OpenAI compatibility instead of native Gemini and are not counted as Gemini wire. Cohere preset similarly uses OpenAI compatibility, not Cohere native chat. No native Cohere or Bedrock ingress conversion is registered. Local recorder and response checks establish 12 pairs. Responses mock token-details correction is required for Messages-to-Responses. Evidence: results/diagnostics/refresh-configuration/agentgateway-protocol-preflight.json | — |
| cohere → gemini | — | — | not_declared: Pinned v1.7.0 image and source 26f5aebe99a36a037fe54f8080a608aca5547e15: CHAT_TRANSLATIONS in crates/agentgateway/src/llm/mod.rs defines supported conversions. Separate custom providers constrain Chat versus Responses wire. Gemini native and Chat-to-Gemini are supported; Messages/Responses use Google OpenAI compatibility instead of native Gemini and are not counted as Gemini wire. Cohere preset similarly uses OpenAI compatibility, not Cohere native chat. No native Cohere or Bedrock ingress conversion is registered. Local recorder and response checks establish 12 pairs. Responses mock token-details correction is required for Messages-to-Responses. Evidence: results/diagnostics/refresh-configuration/agentgateway-protocol-preflight.json | — |
| gemini → gemini | added_latency_p50_us | 211 | valid; gateway_c1_samples=25203; direct_c1_samples=241373 | — |
| gemini → gemini | added_latency_p99_us | 248 | valid; gateway_c1_samples=25203; direct_c1_samples=241373 | — |
| gemini → gemini | gateway_c1_p99_us | 278 | valid; gateway_c1_samples=25203 | — |
| gemini → gemini | direct_c1_p99_us | 30 | valid; direct_c1_samples=241373 | — |
| gemini → gemini | cpu_us_per_request | 196.89006971107798 | valid | — |
| gemini → gemini | rps_per_cpu_second | 5078.976311336717 | valid | — |
| gemini → gemini | cost_window_conc | 8 | valid | — |
| gemini → gemini | cost_core_utilisation | 0.9820473628139751 | valid | — |
| gemini → gemini | cost_window_ok | 120067 | valid | — |
| gemini → gemini | cost_window_rps | 20008 | valid | — |
| gemini → gemini | cost_threads | 13 | valid | — |
| gemini → gemini | cost_nonvol_ctxt_per_request | 0 | valid | — |
| gemini → gemini | cost_majflt | 0 | valid | — |
| gemini → gemini | added_ttft_p50_us | untestable | untestable | — |
| gemini → gemini | added_ttft_p99_us | untestable | untestable | — |
| gemini → gemini | added_gap_p50_us | untestable | untestable | — |
| gemini → gemini | added_gap_p99_us | untestable | untestable | — |
| gemini → gemini | streams_sustained_fps | untestable | untestable | — |
| gemini → gemini | streams_sustained | untestable | untestable | — |
| gemini → gemini | idle_rss_mib | 33.95703125 | valid | — |
| gemini → gemini | steady_state_rss_mib | 54.23828125 | valid | — |
| gemini → gemini | recovered_rss_mib | 50.96484375 | valid | — |
| gemini → gemini | peak_rss_mib | 54.76953125 | valid | — |
| gemini → gemini | peak_rss_hwm_mib | 54.5234375 | valid | — |
| gemini → gemini | time_to_plateau_s | not_measured | not_measured | — |
| gemini → gemini | growth_rate_mib_per_min | -0.1775921183649501 | valid | — |
| gemini → gemini | load_s | 300 | valid | — |
| gemini → gemini | shape | 0 | valid | — |
| gemini → gemini | idle_shape | not_measured | not_measured | — |
| gemini → gemini | idle_static | 1 | valid | — |
| gemini → gemini | idle_growth_rate_mib_per_min | 0.041368822953469664 | valid | — |
| gemini → gemini | throughput_rps | 19915 req/s | concurrency 8; samples 119508 | 1 |
| gemini → gemini | throughput_rps | 20912 req/s | concurrency 64; samples 125481 | 5 |
| gemini → gemini | throughput_rps | 20976 req/s | concurrency 128; samples 125877 | 10 |
| gemini → gemini | throughput_rps | 21168 req/s | concurrency 256; samples 127023 | 50 |
| gemini → gemini | throughput_rps | 21168 req/s | concurrency 256; samples 127023 | 100 |
| gemini → gemini | throughput_rps | 21168 req/s | concurrency 256; samples 127023 | unbounded |
| openai-responses → gemini | — | — | not_declared: Pinned v1.7.0 image and source 26f5aebe99a36a037fe54f8080a608aca5547e15: CHAT_TRANSLATIONS in crates/agentgateway/src/llm/mod.rs defines supported conversions. Separate custom providers constrain Chat versus Responses wire. Gemini native and Chat-to-Gemini are supported; Messages/Responses use Google OpenAI compatibility instead of native Gemini and are not counted as Gemini wire. Cohere preset similarly uses OpenAI compatibility, not Cohere native chat. No native Cohere or Bedrock ingress conversion is registered. Local recorder and response checks establish 12 pairs. Responses mock token-details correction is required for Messages-to-Responses. Evidence: results/diagnostics/refresh-configuration/agentgateway-protocol-preflight.json | — |
| openai → gemini | added_latency_p50_us | 217 | valid; gateway_c1_samples=24639; direct_c1_samples=244550 | — |
| openai → gemini | added_latency_p99_us | 253 | valid; gateway_c1_samples=24639; direct_c1_samples=244550 | — |
| openai → gemini | gateway_c1_p99_us | 283 | valid; gateway_c1_samples=24639 | — |
| openai → gemini | direct_c1_p99_us | 30 | valid; direct_c1_samples=244550 | — |
| openai → gemini | cpu_us_per_request | 202.1159412864718 | valid | — |
| openai → gemini | rps_per_cpu_second | 4947.65525982256 | valid | — |
| openai → gemini | cost_window_conc | 8 | valid | — |
| openai → gemini | cost_core_utilisation | 0.9832975723418331 | valid | — |
| openai → gemini | cost_window_ok | 117111 | valid | — |
| openai → gemini | cost_window_rps | 19516 | valid | — |
| openai → gemini | cost_threads | 13 | valid | — |
| openai → gemini | cost_nonvol_ctxt_per_request | 0 | valid | — |
| openai → gemini | cost_majflt | 0 | valid | — |
| openai → gemini | added_ttft_p50_us | untestable | untestable | — |
| openai → gemini | added_ttft_p99_us | untestable | untestable | — |
| openai → gemini | added_gap_p50_us | untestable | untestable | — |
| openai → gemini | added_gap_p99_us | untestable | untestable | — |
| openai → gemini | streams_sustained_fps | untestable | untestable | — |
| openai → gemini | streams_sustained | untestable | untestable | — |
| openai → gemini | idle_rss_mib | 34.05859375 | valid | — |
| openai → gemini | steady_state_rss_mib | 53.58984375 | valid | — |
| openai → gemini | recovered_rss_mib | 50.46875 | valid | — |
| openai → gemini | peak_rss_mib | 54.25 | valid | — |
| openai → gemini | peak_rss_hwm_mib | 53.984375 | valid | — |
| openai → gemini | time_to_plateau_s | not_measured | not_measured | — |
| openai → gemini | growth_rate_mib_per_min | -0.36093857733091694 | valid | — |
| openai → gemini | load_s | 300 | valid | — |
| openai → gemini | shape | 0 | valid | — |
| openai → gemini | idle_shape | not_measured | not_measured | — |
| openai → gemini | idle_static | 1 | valid | — |
| openai → gemini | idle_growth_rate_mib_per_min | 0.0024901789418739434 | valid | — |
| openai → gemini | throughput_rps | 19532 req/s | concurrency 8; samples 117207 | 1 |
| openai → gemini | throughput_rps | 20455 req/s | concurrency 64; samples 122740 | 5 |
| openai → gemini | throughput_rps | 20455 req/s | concurrency 64; samples 122740 | 10 |
| openai → gemini | throughput_rps | 20723 req/s | concurrency 256; samples 124350 | 50 |
| openai → gemini | throughput_rps | 20723 req/s | concurrency 256; samples 124350 | 100 |
| openai → gemini | throughput_rps | 20723 req/s | concurrency 256; samples 124350 | unbounded |
| bedrock → gemini | — | — | not_declared: Pinned v1.7.0 image and source 26f5aebe99a36a037fe54f8080a608aca5547e15: CHAT_TRANSLATIONS in crates/agentgateway/src/llm/mod.rs defines supported conversions. Separate custom providers constrain Chat versus Responses wire. Gemini native and Chat-to-Gemini are supported; Messages/Responses use Google OpenAI compatibility instead of native Gemini and are not counted as Gemini wire. Cohere preset similarly uses OpenAI compatibility, not Cohere native chat. No native Cohere or Bedrock ingress conversion is registered. Local recorder and response checks establish 12 pairs. Responses mock token-details correction is required for Messages-to-Responses. Evidence: results/diagnostics/refresh-configuration/agentgateway-protocol-preflight.json | — |
| anthropic → gemini | — | — | not_declared: Pinned v1.7.0 image and source 26f5aebe99a36a037fe54f8080a608aca5547e15: CHAT_TRANSLATIONS in crates/agentgateway/src/llm/mod.rs defines supported conversions. Separate custom providers constrain Chat versus Responses wire. Gemini native and Chat-to-Gemini are supported; Messages/Responses use Google OpenAI compatibility instead of native Gemini and are not counted as Gemini wire. Cohere preset similarly uses OpenAI compatibility, not Cohere native chat. No native Cohere or Bedrock ingress conversion is registered. Local recorder and response checks establish 12 pairs. Responses mock token-details correction is required for Messages-to-Responses. Evidence: results/diagnostics/refresh-configuration/agentgateway-protocol-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 |
| --- | --- | --- | --- | --- | --- | --- | --- |
| openai-responses → openai-responses | served | — | — | — | Recorded run evidence | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| 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) | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| 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) | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| anthropic → openai-responses | served | — | — | — | Recorded run evidence | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| 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) | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| 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) | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| anthropic → bedrock | served | — | — | — | Recorded run evidence | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| 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) | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| 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) | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| openai-responses → bedrock | served | — | — | — | Recorded run evidence | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| bedrock → 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) | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| openai → bedrock | served | — | — | — | Recorded run evidence | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| anthropic → 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) | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| 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) | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| 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) | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| cohere → 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) | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| openai → 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) | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| 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) | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| openai-responses → openai | served | — | — | — | Recorded run evidence | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| 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) | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| openai → openai | served | — | — | — | Recorded run evidence | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| anthropic → openai | served | — | — | — | Recorded run evidence | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| 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) | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| 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) | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| anthropic → anthropic | served | — | — | — | Recorded run evidence | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| 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) | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| 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) | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| openai → anthropic | served | — | — | — | Recorded run evidence | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| 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) | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| 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) | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| 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) | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| gemini → gemini | served | — | — | — | Recorded run evidence | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| 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) | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| openai → gemini | served | — | — | — | Recorded run evidence | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| 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) | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |
| anthropic → 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) | — | ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b |

**Harness commit**: `d240badd966933fa19eec0606b2d1f3945f18ad2`

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

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

**somaxconn**: 65535

**tcp_max_syn_backlog**: 65535

## Stream sweep diagnostics

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

| Ingress → upstream | Concurrency | Lanes by HTTP status | Lanes with no response |
| --- | --- | --- | --- |
| openai → openai | 1 | {"200":1} | 0 |
| openai → openai | 2 | {"200":2} | 0 |
| openai → openai | 4 | {"200":4} | 0 |
| openai → openai | 8 | {"200":8} | 0 |
| openai → openai | 16 | {"200":16} | 0 |
| openai → openai | 32 | {"200":32} | 0 |
| openai → openai | 64 | {"200":64} | 0 |
| openai → openai | 128 | {"200":128} | 0 |
| openai → openai | 256 | {"200":256} | 0 |
| openai → openai | 512 | {"200":512} | 0 |
| openai → openai | 1024 | {"200":1024} | 0 |
| openai → openai | 2048 | {"200":2048} | 0 |
| openai → openai | 4096 | {"200":4096} | 0 |
| openai → openai | 3072 | {"200":3072} | 0 |
| openai → openai | 2560 | {"200":2560} | 0 |
| openai → openai | 2304 | {"200":2304} | 0 |
| openai → openai | 2432 | {"200":2432} | 0 |
| openai → openai | 2496 | {"200":2496} | 0 |
| openai → openai | 2528 | {"200":2528} | 0 |
| openai → openai | 2544 | {"200":2544} | 0 |
| openai → openai | 2552 | {"200":2552} | 0 |
| openai → openai | 2556 | {"200":2556} | 0 |
| openai → openai | 2558 | {"200":2558} | 0 |
| openai → openai | 2557 | {"200":2557} | 0 |
| openai → openai | 2556 | {"200":2556} | 0 |
| openai → openai | 2556 | {"200":2556} | 0 |
| openai → openai | 2302 | {"200":2302} | 0 |
| openai → openai | 2175 | {"200":2175} | 0 |
| openai → openai | 2111 | {"200":2111} | 0 |
| openai → openai | 2079 | {"200":2079} | 0 |
| openai → openai | 2048 | {"200":2048} | 0 |
| openai → openai | 2048 | {"200":2048} | 0 |
| openai → openai | 2048 | {"200":2048} | 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":8192} | 0 |
| anthropic → openai | 6144 | {"200":6144} | 0 |
| anthropic → openai | 5120 | {"200":5120} | 0 |
| anthropic → openai | 4608 | {"200":4608} | 0 |
| anthropic → openai | 4352 | {"200":4352} | 0 |
| anthropic → openai | 4224 | {"200":4224} | 0 |
| anthropic → openai | 4160 | {"200":4160} | 0 |
| anthropic → openai | 4128 | {"200":4128} | 0 |
| anthropic → openai | 4144 | {"200":4144} | 0 |
| anthropic → openai | 4136 | {"200":3945} | 191 |
| anthropic → openai | 4132 | {"200":3959} | 173 |
| anthropic → openai | 4130 | {"200":4130} | 0 |
| anthropic → openai | 4129 | {"200":4129} | 0 |
| anthropic → openai | 4128 | {"200":3234} | 894 |
| anthropic → openai | 4128 | {"200":4057} | 71 |
| anthropic → openai | 4112 | {"200":3242} | 870 |
| anthropic → openai | 4104 | {"200":2957} | 1147 |
| anthropic → openai | 4100 | {"200":3860} | 240 |
| anthropic → openai | 4098 | {"200":2704} | 1394 |
| anthropic → openai | 4096 | {"200":3404} | 692 |
| anthropic → openai | 4096 | {"200":3208} | 888 |
| anthropic → openai | 4096 | {"200":3250} | 846 |
| 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 | 6144 | {"200":6144} | 0 |
| anthropic → anthropic | 5120 | {"200":5120} | 0 |
| anthropic → anthropic | 4608 | {"200":4608} | 0 |
| anthropic → anthropic | 4352 | {"200":4352} | 0 |
| anthropic → anthropic | 4224 | {"200":4224} | 0 |
| anthropic → anthropic | 4160 | {"200":4160} | 0 |
| anthropic → anthropic | 4128 | {"200":4128} | 0 |
| anthropic → anthropic | 4112 | {"200":4112} | 0 |
| anthropic → anthropic | 4104 | {"200":4104} | 0 |
| anthropic → anthropic | 4108 | {"200":4108} | 0 |
| anthropic → anthropic | 4106 | {"200":4106} | 0 |
| anthropic → anthropic | 4105 | {"200":4105} | 0 |
| anthropic → anthropic | 4104 | {"200":4104} | 0 |
| anthropic → anthropic | 4104 | {"200":3923} | 181 |
| 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 | 3072 | {"200":3072} | 0 |
| openai → anthropic | 2560 | {"200":2560} | 0 |
| openai → anthropic | 2304 | {"200":2304} | 0 |
| openai → anthropic | 2432 | {"200":2432} | 0 |
| openai → anthropic | 2496 | {"200":2496} | 0 |
| openai → anthropic | 2528 | {"200":2528} | 0 |
| openai → anthropic | 2544 | {"200":2544} | 0 |
| openai → anthropic | 2552 | {"200":2552} | 0 |
| openai → anthropic | 2556 | {"200":2556} | 0 |
| openai → anthropic | 2558 | {"200":2558} | 0 |
| openai → anthropic | 2559 | {"200":2559} | 0 |
| openai → anthropic | 2559 | {"200":2559} | 0 |
| openai → anthropic | 2559 | {"200":2559} | 0 |

## Recorded notes

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

- Pinned v1.7.0 image and source 26f5aebe99a36a037fe54f8080a608aca5547e15: CHAT_TRANSLATIONS in crates/agentgateway/src/llm/mod.rs defines supported conversions. Separate custom providers constrain Chat versus Responses wire. Gemini native and Chat-to-Gemini are supported; Messages/Responses use Google OpenAI compatibility instead of native Gemini and are not counted as Gemini wire. Cohere preset similarly uses OpenAI compatibility, not Cohere native chat. No native Cohere or Bedrock ingress conversion is registered. Local recorder and response checks establish 12 pairs. Responses mock token-details correction is required for Messages-to-Responses. Evidence: results/diagnostics/refresh-configuration/agentgateway-protocol-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-agentgateway-fa351efdcee949e977dd1d4331440512b3e913a74b878f4bd1b19a7c7ca2900c](https://onthebench.ai/gateways/runs/run-agentgateway-fa351efdcee949e977dd1d4331440512b3e913a74b878f4bd1b19a7c7ca2900c)  ·  [complete raw record including RSS series](https://onthebench.ai/api/v1/runs/run-agentgateway-fa351efdcee949e977dd1d4331440512b3e913a74b878f4bd1b19a7c7ca2900c/raw.json)

## Configuration recorded by the run

```
llm:
  port: 8080
  models:
  - name: gpt-4o-mini
    provider: openAI
    params:
      model: gpt-4o-mini
      apiKey: "sk-dummy-openai"
      baseUrl: "http://127.0.0.1:8000/v1"
  - name: claude-3-5-sonnet-20241022
    provider: anthropic
    params:
      model: claude-3-5-sonnet-20241022
      apiKey: "sk-ant-dummy"
      baseUrl: "http://127.0.0.1:8000/v1"
  - name: anthropic.claude-3-sonnet-20240229-v1:0
    provider: bedrock
    params:
      model: anthropic.claude-3-sonnet-20240229-v1:0
      awsRegion: us-east-1
      baseUrl: "http://127.0.0.1:8000/bedrock"
  - name: gemini-1.5-pro
    provider: gemini
    params:
      model: gemini-1.5-pro
      apiKey: dummy
      baseUrl: http://127.0.0.1:8000
  - name: bench-chat
    provider:
      custom:
        formats:
        - type: completions
    params:
      model: gpt-4o-mini
      apiKey: dummy
      baseUrl: http://127.0.0.1:8000/v1
  - name: bench-responses
    provider:
      custom:
        formats:
        - type: responses
    params:
      model: gpt-4o-mini
      apiKey: dummy
      baseUrl: http://127.0.0.1:8000/v1

```

## Compare a shared protocol cell

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

## Historical measurements

- [2026-11-10T17:39:19Z — ghcr.io/agentgateway/agentgateway:v1.7.0@sha256:4562330cd6e580bfac2de5b8905fee608dcf3fff6f1f4900744d27bdddb3b41b](https://onthebench.ai/gateways/runs/run-agentgateway-fa351efdcee949e977dd1d4331440512b3e913a74b878f4bd1b19a7c7ca2900c)
- [2026-08-26T19:08:42Z — ghcr.io/agentgateway/agentgateway:v1.4.1](https://onthebench.ai/gateways/runs/run-agentgateway-b908faf77122520df0d8055d83ae1c95022a63e5302463fa419aa5d7c3696894)
