# One-API benchmark measurements

Operated by the Busbar team; Busbar is also an entrant. Gateway overhead against a local mock is not end-to-end model latency.  [Methodology and limitations](https://onthebench.ai/gateways/method) .

Go · recorded build: justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477. Measured: 2026-11-10T17:39:58Z.

[JSON](https://onthebench.ai/api/v1/gateways/one-api)  ·  [Markdown](https://onthebench.ai/gateways/one-api.md)  ·  [CSV](https://onthebench.ai/gateways/data.csv?gateway=one-api&compact=1)  ·  [Project source](https://github.com/songquanpeng/one-api)  ·  [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) |
| --- | --- | --- | --- | --- |
| anthropic → openai-responses | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| openai → openai-responses | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| gemini → openai-responses | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| cohere → openai-responses | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| openai-responses → openai-responses | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| bedrock → openai-responses | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| openai-responses → gemini | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| cohere → gemini | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| bedrock → gemini | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| anthropic → gemini | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| gemini → gemini | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| openai → gemini | added_latency_p50_us | 34189 | valid; gateway_c1_samples=178; direct_c1_samples=213320 | — |
| openai → gemini | added_latency_p99_us | 34903 | valid; gateway_c1_samples=178; direct_c1_samples=213320 | — |
| openai → gemini | gateway_c1_p99_us | 34941 | valid; gateway_c1_samples=178 | — |
| openai → gemini | direct_c1_p99_us | 38 | valid; direct_c1_samples=213320 | — |
| openai → gemini | cpu_us_per_request | 3395.3488372093025 | valid | — |
| openai → gemini | rps_per_cpu_second | 294.52054794520546 | valid | — |
| openai → gemini | cost_window_conc | 8 | valid | — |
| openai → gemini | cost_core_utilisation | 0.02833912022723337 | valid | — |
| openai → gemini | cost_window_ok | 215 | valid | — |
| openai → gemini | cost_window_rps | 35 | valid | — |
| openai → gemini | cost_threads | 84 | 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 | 80.31640625 | valid | — |
| openai → gemini | steady_state_rss_mib | 144.08203125 | valid | — |
| openai → gemini | recovered_rss_mib | 141.51171875 | valid | — |
| openai → gemini | peak_rss_mib | 144.8203125 | valid | — |
| openai → gemini | peak_rss_hwm_mib | 144.71875 | valid | — |
| openai → gemini | time_to_plateau_s | 120 | valid | — |
| openai → gemini | growth_rate_mib_per_min | 0.4419008197056353 | valid | — |
| openai → gemini | load_s | 300 | valid | — |
| openai → gemini | shape | not_measured | not_measured | — |
| openai → gemini | idle_shape | not_measured | not_measured | — |
| openai → gemini | idle_static | 1 | valid | — |
| openai → gemini | idle_growth_rate_mib_per_min | -0.1152909126823189 | valid | — |
| openai → gemini | throughput_rps | bound_not_met req/s | concurrency not recorded; every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 |
| openai → gemini | throughput_rps | bound_not_met req/s | concurrency not recorded; every cleanly-served rung that reported a tail had it at or above 5ms, so this gateway carried no measurable throughput under that bound | 5 |
| openai → gemini | throughput_rps | bound_not_met req/s | concurrency not recorded; every cleanly-served rung that reported a tail had it at or above 10ms, so this gateway carried no measurable throughput under that bound | 10 |
| openai → gemini | throughput_rps | 30 req/s | concurrency 1; samples 181 | 50 |
| openai → gemini | throughput_rps | 30 req/s | concurrency 1; samples 181 | 100 |
| openai → gemini | throughput_rps | 38 req/s | concurrency 16; samples 230 | unbounded |
| bedrock → cohere | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| gemini → cohere | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| anthropic → cohere | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| cohere → cohere | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| openai-responses → cohere | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| openai → cohere | added_latency_p50_us | 34053 | valid; gateway_c1_samples=178; direct_c1_samples=265039 | — |
| openai → cohere | added_latency_p99_us | 42953 | valid; gateway_c1_samples=178; direct_c1_samples=265039 | — |
| openai → cohere | gateway_c1_p99_us | 42981 | valid; gateway_c1_samples=178 | — |
| openai → cohere | direct_c1_p99_us | 28 | valid; direct_c1_samples=265039 | — |
| openai → cohere | cpu_us_per_request | 3063.063063063063 | valid | — |
| openai → cohere | rps_per_cpu_second | 326.4705882352941 | valid | — |
| openai → cohere | cost_window_conc | 8 | valid | — |
| openai → cohere | cost_core_utilisation | 0.02686924016940405 | valid | — |
| openai → cohere | cost_window_ok | 222 | valid | — |
| openai → cohere | cost_window_rps | 36 | valid | — |
| openai → cohere | cost_threads | 79 | valid | — |
| openai → cohere | cost_nonvol_ctxt_per_request | 0 | valid | — |
| openai → cohere | cost_majflt | 0 | valid | — |
| openai → cohere | added_ttft_p50_us | untestable | untestable | — |
| openai → cohere | added_ttft_p99_us | untestable | untestable | — |
| openai → cohere | added_gap_p50_us | untestable | untestable | — |
| openai → cohere | added_gap_p99_us | untestable | untestable | — |
| openai → cohere | streams_sustained_fps | untestable | untestable | — |
| openai → cohere | streams_sustained | untestable | untestable | — |
| openai → cohere | idle_rss_mib | 86.46875 | valid | — |
| openai → cohere | steady_state_rss_mib | 141.58984375 | valid | — |
| openai → cohere | recovered_rss_mib | 134.26953125 | valid | — |
| openai → cohere | peak_rss_mib | 142.06640625 | valid | — |
| openai → cohere | peak_rss_hwm_mib | 142.01953125 | valid | — |
| openai → cohere | time_to_plateau_s | 127 | valid | — |
| openai → cohere | growth_rate_mib_per_min | 0.8110074898737798 | valid | — |
| openai → cohere | load_s | 307 | valid | — |
| openai → cohere | shape | not_measured | not_measured | — |
| openai → cohere | idle_shape | not_measured | not_measured | — |
| openai → cohere | idle_static | 1 | valid | — |
| openai → cohere | idle_growth_rate_mib_per_min | -0.11088631518992631 | valid | — |
| openai → cohere | throughput_rps | bound_not_met req/s | concurrency not recorded; every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 |
| openai → cohere | throughput_rps | bound_not_met req/s | concurrency not recorded; every cleanly-served rung that reported a tail had it at or above 5ms, so this gateway carried no measurable throughput under that bound | 5 |
| openai → cohere | throughput_rps | bound_not_met req/s | concurrency not recorded; every cleanly-served rung that reported a tail had it at or above 10ms, so this gateway carried no measurable throughput under that bound | 10 |
| openai → cohere | throughput_rps | 29 req/s | concurrency 1; samples 178 | 50 |
| openai → cohere | throughput_rps | 29 req/s | concurrency 1; samples 178 | 100 |
| openai → cohere | throughput_rps | 36 req/s | concurrency 8; samples 220 | unbounded |
| openai → bedrock (upstream over HTTPS) | added_latency_p50_us | 35450 | valid; gateway_c1_samples=170; direct_c1_samples=258581 | — |
| openai → bedrock (upstream over HTTPS) | added_latency_p99_us | 35742 | valid; gateway_c1_samples=170; direct_c1_samples=258581 | — |
| openai → bedrock (upstream over HTTPS) | gateway_c1_p99_us | 35771 | valid; gateway_c1_samples=170 | — |
| openai → bedrock (upstream over HTTPS) | direct_c1_p99_us | 29 | valid; direct_c1_samples=258581 | — |
| openai → bedrock (upstream over HTTPS) | cpu_us_per_request | 4976.076555023923 | valid | — |
| openai → bedrock (upstream over HTTPS) | rps_per_cpu_second | 200.96153846153845 | valid | — |
| openai → bedrock (upstream over HTTPS) | cost_window_conc | 8 | valid | — |
| openai → bedrock (upstream over HTTPS) | cost_core_utilisation | 0.040498520437646256 | valid | — |
| openai → bedrock (upstream over HTTPS) | cost_window_ok | 209 | valid | — |
| openai → bedrock (upstream over HTTPS) | cost_window_rps | 34 | valid | — |
| openai → bedrock (upstream over HTTPS) | cost_threads | 89 | valid | — |
| openai → bedrock (upstream over HTTPS) | cost_nonvol_ctxt_per_request | 0 | valid | — |
| openai → bedrock (upstream over HTTPS) | cost_majflt | 0 | valid | — |
| openai → bedrock (upstream over HTTPS) | added_ttft_p50_us | untestable | untestable | — |
| openai → bedrock (upstream over HTTPS) | added_ttft_p99_us | untestable | untestable | — |
| openai → bedrock (upstream over HTTPS) | added_gap_p50_us | untestable | untestable | — |
| openai → bedrock (upstream over HTTPS) | added_gap_p99_us | untestable | untestable | — |
| openai → bedrock (upstream over HTTPS) | streams_sustained_fps | untestable | untestable | — |
| openai → bedrock (upstream over HTTPS) | streams_sustained | untestable | untestable | — |
| openai → bedrock (upstream over HTTPS) | idle_rss_mib | 85.7578125 | valid | — |
| openai → bedrock (upstream over HTTPS) | steady_state_rss_mib | 348.8046875 | valid | — |
| openai → bedrock (upstream over HTTPS) | recovered_rss_mib | 332.67578125 | valid | — |
| openai → bedrock (upstream over HTTPS) | peak_rss_mib | 349.359375 | valid | — |
| openai → bedrock (upstream over HTTPS) | peak_rss_hwm_mib | 349.30859375 | valid | — |
| openai → bedrock (upstream over HTTPS) | time_to_plateau_s | 174 | valid | — |
| openai → bedrock (upstream over HTTPS) | growth_rate_mib_per_min | 0.6047438141000504 | valid | — |
| openai → bedrock (upstream over HTTPS) | load_s | 303 | valid | — |
| openai → bedrock (upstream over HTTPS) | shape | not_measured | not_measured | — |
| openai → bedrock (upstream over HTTPS) | idle_shape | not_measured | not_measured | — |
| openai → bedrock (upstream over HTTPS) | idle_static | 1 | valid | — |
| openai → bedrock (upstream over HTTPS) | idle_growth_rate_mib_per_min | -0.11095193717665752 | valid | — |
| openai → bedrock (upstream over HTTPS) | throughput_rps | bound_not_met req/s | concurrency not recorded; every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 |
| openai → bedrock (upstream over HTTPS) | throughput_rps | bound_not_met req/s | concurrency not recorded; every cleanly-served rung that reported a tail had it at or above 5ms, so this gateway carried no measurable throughput under that bound | 5 |
| openai → bedrock (upstream over HTTPS) | throughput_rps | bound_not_met req/s | concurrency not recorded; every cleanly-served rung that reported a tail had it at or above 10ms, so this gateway carried no measurable throughput under that bound | 10 |
| openai → bedrock (upstream over HTTPS) | throughput_rps | 28 req/s | concurrency 1; samples 172 | 50 |
| openai → bedrock (upstream over HTTPS) | throughput_rps | 28 req/s | concurrency 1; samples 172 | 100 |
| openai → bedrock (upstream over HTTPS) | throughput_rps | 34 req/s | concurrency 8; samples 209 | unbounded |
| cohere → bedrock | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| gemini → bedrock | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| anthropic → bedrock | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| bedrock → bedrock | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| openai-responses → bedrock | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| openai → anthropic | added_latency_p50_us | 34189 | valid; gateway_c1_samples=97; direct_c1_samples=264624 | — |
| openai → anthropic | added_latency_p99_us | 1029936 | valid; gateway_c1_samples=97; direct_c1_samples=264624 | — |
| openai → anthropic | gateway_c1_p99_us | 1029964 | valid; gateway_c1_samples=97 | — |
| openai → anthropic | direct_c1_p99_us | 28 | valid; direct_c1_samples=264624 | — |
| openai → anthropic | cpu_us_per_request | 3348.8372093023254 | valid | — |
| openai → anthropic | rps_per_cpu_second | 298.61111111111114 | valid | — |
| openai → anthropic | cost_window_conc | 8 | valid | — |
| openai → anthropic | cost_core_utilisation | 0.02727559585540872 | valid | — |
| openai → anthropic | cost_window_ok | 215 | valid | — |
| openai → anthropic | cost_window_rps | 35 | valid | — |
| openai → anthropic | cost_threads | 214 | valid | — |
| openai → anthropic | cost_nonvol_ctxt_per_request | 0 | valid | — |
| openai → anthropic | cost_majflt | 0 | valid | — |
| openai → anthropic | added_ttft_p50_us | 751 | valid; ttft_gw_samples=100; ttft_direct_samples=100 | — |
| openai → anthropic | added_ttft_p99_us | 1758 | valid; ttft_gw_samples=100; ttft_direct_samples=100 | — |
| openai → anthropic | added_gap_p50_us | 0 | valid; gap_gw_samples=63; gap_direct_samples=63 | — |
| openai → anthropic | added_gap_p99_us | 7 | valid; gap_gw_samples=63; gap_direct_samples=63 | — |
| openai → anthropic | streams_sustained_fps | 4429.711949178542 | valid | — |
| openai → anthropic | streams_sustained | 106 | valid | — |
| openai → anthropic | idle_rss_mib | 77.828125 | valid | — |
| openai → anthropic | steady_state_rss_mib | 142.8125 | valid | — |
| openai → anthropic | recovered_rss_mib | 139.71484375 | valid | — |
| openai → anthropic | peak_rss_mib | 143.05859375 | valid | — |
| openai → anthropic | peak_rss_hwm_mib | 143.05078125 | valid | — |
| openai → anthropic | time_to_plateau_s | 129 | valid | — |
| openai → anthropic | growth_rate_mib_per_min | 0.17067641454619376 | valid | — |
| openai → anthropic | load_s | 301 | valid | — |
| openai → anthropic | shape | not_measured | not_measured | — |
| openai → anthropic | idle_shape | not_measured | not_measured | — |
| openai → anthropic | idle_static | 1 | valid | — |
| openai → anthropic | idle_growth_rate_mib_per_min | -0.1148098714721732 | valid | — |
| openai → anthropic | throughput_rps | bound_not_met req/s | concurrency not recorded; every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 |
| openai → anthropic | throughput_rps | bound_not_met req/s | concurrency not recorded; every cleanly-served rung that reported a tail had it at or above 5ms, so this gateway carried no measurable throughput under that bound | 5 |
| openai → anthropic | throughput_rps | bound_not_met req/s | concurrency not recorded; every cleanly-served rung that reported a tail had it at or above 10ms, so this gateway carried no measurable throughput under that bound | 10 |
| openai → anthropic | throughput_rps | 30 req/s | concurrency 1; samples 182 | 50 |
| openai → anthropic | throughput_rps | 30 req/s | concurrency 1; samples 182 | 100 |
| openai → anthropic | throughput_rps | 38 req/s | concurrency 16; samples 229 | unbounded |
| openai-responses → anthropic | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| cohere → anthropic | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| gemini → anthropic | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| bedrock → anthropic | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| anthropic → anthropic | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| openai-responses → openai | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| cohere → openai | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| openai → openai | added_latency_p50_us | 34017 | valid; gateway_c1_samples=102; direct_c1_samples=263027 | — |
| openai → openai | added_latency_p99_us | 589609 | valid; gateway_c1_samples=102; direct_c1_samples=263027 | — |
| openai → openai | gateway_c1_p99_us | 589637 | valid; gateway_c1_samples=102 | — |
| openai → openai | direct_c1_p99_us | 28 | valid; direct_c1_samples=263027 | — |
| openai → openai | cpu_us_per_request | not_measured | not_measured | — |
| openai → openai | rps_per_cpu_second | not_measured | not_measured | — |
| openai → openai | cost_window_conc | not_measured | not_measured | — |
| openai → openai | cost_core_utilisation | not_measured | not_measured | — |
| openai → openai | cost_window_ok | not_measured | not_measured | — |
| openai → openai | cost_window_rps | not_measured | not_measured | — |
| openai → openai | cost_threads | not_measured | not_measured | — |
| openai → openai | cost_nonvol_ctxt_per_request | not_measured | not_measured | — |
| openai → openai | cost_majflt | not_measured | not_measured | — |
| openai → openai | added_ttft_p50_us | 728 | valid; ttft_gw_samples=100; ttft_direct_samples=100 | — |
| openai → openai | added_ttft_p99_us | 770 | valid; ttft_gw_samples=100; ttft_direct_samples=100 | — |
| openai → openai | added_gap_p50_us | 12 | valid; gap_gw_samples=63; gap_direct_samples=63 | — |
| openai → openai | added_gap_p99_us | 27 | valid; gap_gw_samples=63; gap_direct_samples=63 | — |
| openai → openai | streams_sustained_fps | 7497.836670339175 | valid | — |
| openai → openai | streams_sustained | 213 | valid | — |
| openai → openai | idle_rss_mib | 87.15625 | valid | — |
| openai → openai | steady_state_rss_mib | 141.876953125 | valid | — |
| openai → openai | recovered_rss_mib | 136.578125 | valid | — |
| openai → openai | peak_rss_mib | 142.4921875 | valid | — |
| openai → openai | peak_rss_hwm_mib | 142.47265625 | valid | — |
| openai → openai | time_to_plateau_s | 134 | valid | — |
| openai → openai | growth_rate_mib_per_min | 0.1159464817427444 | valid | — |
| openai → openai | load_s | 300 | valid | — |
| openai → openai | shape | not_measured | not_measured | — |
| openai → openai | idle_shape | not_measured | not_measured | — |
| openai → openai | idle_static | 1 | valid | — |
| openai → openai | idle_growth_rate_mib_per_min | -0.1148904573303369 | valid | — |
| openai → openai | throughput_rps | bound_not_met req/s | concurrency not recorded; every cleanly-served rung that reported a tail had it at or above 1ms, so this gateway carried no measurable throughput under that bound | 1 |
| openai → openai | throughput_rps | bound_not_met req/s | concurrency not recorded; every cleanly-served rung that reported a tail had it at or above 5ms, so this gateway carried no measurable throughput under that bound | 5 |
| openai → openai | throughput_rps | bound_not_met req/s | concurrency not recorded; every cleanly-served rung that reported a tail had it at or above 10ms, so this gateway carried no measurable throughput under that bound | 10 |
| openai → openai | throughput_rps | 32 req/s | concurrency 1; samples 196 | 50 |
| openai → openai | throughput_rps | 32 req/s | concurrency 1; samples 196 | 100 |
| openai → openai | throughput_rps | 38 req/s | concurrency 16; samples 229 | unbounded |
| anthropic → openai | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| gemini → openai | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |
| bedrock → openai | — | — | not_declared: Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go | — |

## 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 |
| --- | --- | --- | --- | --- | --- | --- | --- |
| anthropic → openai-responses | not_declared | — | [recorded note 1](https://onthebench.ai/#note-1) | [recorded note 1](https://onthebench.ai/#note-1) | [recorded note 2](https://onthebench.ai/#note-2) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| 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) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| 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) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| 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) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| openai-responses → 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) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| 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) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| 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) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| 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) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| 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) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| 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) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| gemini → 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) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| openai → gemini | served | — | — | — | Recorded run evidence | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| 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) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| 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) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| 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) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| 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) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| 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) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| openai → cohere | served | — | — | — | Recorded run evidence | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| openai → bedrock | served | — | — | — | Recorded run evidence | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| 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) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| 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) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| anthropic → bedrock | not_declared | — | [recorded note 1](https://onthebench.ai/#note-1) | [recorded note 1](https://onthebench.ai/#note-1) | [recorded note 2](https://onthebench.ai/#note-2) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| 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) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| openai-responses → bedrock | not_declared | — | [recorded note 1](https://onthebench.ai/#note-1) | [recorded note 1](https://onthebench.ai/#note-1) | [recorded note 2](https://onthebench.ai/#note-2) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| openai → anthropic | served | — | — | — | Recorded run evidence | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| 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) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| 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) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| 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) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| 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) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| anthropic → 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) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| openai-responses → openai | not_declared | — | [recorded note 1](https://onthebench.ai/#note-1) | [recorded note 1](https://onthebench.ai/#note-1) | [recorded note 2](https://onthebench.ai/#note-2) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| 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) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| openai → openai | served | — | — | — | Recorded run evidence | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| anthropic → 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) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| 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) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |
| 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) | — | justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477 |

**Harness commit**: `d240badd966933fa19eec0606b2d1f3945f18ad2`

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

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

**somaxconn**: 65535

**tcp_max_syn_backlog**: 65535

## Stream sweep diagnostics

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

| Ingress → upstream | Concurrency | Lanes by HTTP status | Lanes with no response |
| --- | --- | --- | --- |
| openai → anthropic | 1 | {"200":1} | 0 |
| openai → anthropic | 2 | {"200":2} | 0 |
| openai → anthropic | 4 | {"200":4} | 0 |
| openai → anthropic | 8 | {"200":8} | 0 |
| openai → anthropic | 16 | {"200":16} | 0 |
| openai → anthropic | 32 | {"200":32} | 0 |
| openai → anthropic | 64 | {"200":64} | 0 |
| openai → anthropic | 128 | {"200":115,"401":5,"500":4,"503":4} | 0 |
| openai → anthropic | 96 | {"200":96} | 0 |
| openai → anthropic | 112 | {"200":108,"401":3,"503":1} | 0 |
| openai → anthropic | 104 | {"200":104} | 0 |
| openai → anthropic | 108 | {"200":100,"401":5,"500":1,"503":2} | 0 |
| openai → anthropic | 106 | {"200":106} | 0 |
| openai → anthropic | 107 | {"200":101,"401":5,"500":1} | 0 |
| openai → anthropic | 106 | {"200":106} | 0 |
| openai → anthropic | 106 | {"200":105,"401":1} | 0 |
| openai → openai | 1 | {"200":1} | 0 |
| openai → openai | 2 | {"200":2} | 0 |
| openai → openai | 4 | {"200":4} | 0 |
| openai → openai | 8 | {"200":8} | 0 |
| openai → openai | 16 | {"200":16} | 0 |
| openai → openai | 32 | {"200":32} | 0 |
| openai → openai | 64 | {"200":64} | 0 |
| openai → openai | 128 | {"200":128} | 0 |
| openai → openai | 256 | {"200":238,"401":8,"500":3,"503":7} | 0 |
| openai → openai | 192 | {"200":192} | 0 |
| openai → openai | 224 | {"200":161,"401":33,"500":17,"503":13} | 0 |
| openai → openai | 208 | {"200":208} | 0 |
| openai → openai | 216 | {"200":150,"401":50,"500":6,"503":10} | 0 |
| openai → openai | 212 | {"200":212} | 0 |
| openai → openai | 214 | {"200":131,"401":40,"500":15,"503":28} | 0 |
| openai → openai | 213 | {"200":213} | 0 |
| openai → openai | 213 | {"200":155,"401":37,"500":7,"503":14} | 0 |
| openai → openai | 213 | {"200":213} | 0 |

## Recorded notes

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

- Pinned v0.6.10 source 3915ce9814b8261a1ab13ed93adec58b463cd75c: router/relay.go exposes OpenAI chat ingress, not native Responses, Anthropic, Gemini, Cohere or Bedrock ingress. relay/adaptor supports OpenAI, Anthropic, Gemini, Cohere (v1/chat) and AWS Bedrock (relay/adaptor/aws, InvokeModel through the AWS SDK at its default HTTPS endpoint, which the engine v1.2 catch-all DNS resolves to the mock) upstreams. This matrix describes the tested protocol routes, not all product features. Source: https://github.com/songquanpeng/one-api/blob/3915ce9814b8261a1ab13ed93adec58b463cd75c/router/relay.go
- 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-one-api-c9427cc70e392f48beb8e68d908f8924a7330748ac291dab1cd902b3f14c542f](https://onthebench.ai/gateways/runs/run-one-api-c9427cc70e392f48beb8e68d908f8924a7330748ac291dab1cd902b3f14c542f)  ·  [complete raw record including RSS series](https://onthebench.ai/api/v1/runs/run-one-api-c9427cc70e392f48beb8e68d908f8924a7330748ac291dab1cd902b3f14c542f/raw.json)

## Configuration recorded by the run

```
Not recorded.
```

## Compare a shared protocol cell

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

## Historical measurements

- [2026-11-10T17:39:58Z — justsong/one-api:v0.6.10@sha256:af6b361630bdf408b86fce7aa8fc131b37ea5177ae866d121ba7c002e773e477](https://onthebench.ai/gateways/runs/run-one-api-c9427cc70e392f48beb8e68d908f8924a7330748ac291dab1cd902b3f14c542f)
- [2026-08-03T22:17:39Z — justsong/one-api:v0.6.10](https://onthebench.ai/gateways/runs/run-one-api-11de2fe9f88cb0feb1f7defdde00f6ac6a769b77cb80b027b2362a0652915073)
