Files
sub2api/backend/internal/service/upstream_response_model_bench_test.go
T
feeeei 75faedda95 计费:fast/priority 按上游响应实际档位只降不升计费
此前 OpenAI service_tier 与 Anthropic speed 的计费档位只看请求侧,上游
因容量把 priority/fast 静默降为 default/standard 时仍按 2x 收费。现在:

- upstreamResponseModelObserver 同时观察响应侧档位:OpenAI 只信终止事件
  与无类型 body 里的 service_tier(response.created 为请求回显,忽略),
  Anthropic 读 usage.speed;查找仅在带 model 的帧上触发
- ResolveBillingServiceTier 只降不升:响应档位比请求便宜才采纳,更贵、
  未知或缺失一律沿用请求侧,与图片尺寸归并、response_model 计费的
  "不得更贵"不变式一致
- usage_logs.service_tier 记录实际计费档位,降级写
  billing.service_tier_downgraded 审计日志;HTTP、WS 及 ws_v2 relay 全覆盖
2026-08-23 16:53:28 +08:00

99 lines
2.7 KiB
Go

package service
import (
"strings"
"testing"
"github.com/tidwall/gjson"
)
var (
upstreamResponseModelBenchmarkSink string
upstreamResponseModelBenchmarkTerminal = []byte(`{"type":"response.completed","response":{"id":"resp_123","model":"gpt-5.5-2026-04-23"},"usage":{"input_tokens":128,"output_tokens":64}}`)
upstreamResponseModelBenchmarkDelta = []byte(`{"type":"response.output_text.delta","delta":"hello"}`)
upstreamResponseModelBenchmarkWrapper = []byte(`{"response":{"response":{"modelVersion":"gemini-3-pro","candidates":[]}}}`)
)
func BenchmarkUpstreamResponseModelOpenAI(b *testing.B) {
tests := []struct {
name string
payload []byte
}{
{name: "terminal_with_model", payload: upstreamResponseModelBenchmarkTerminal},
{name: "delta_without_model", payload: upstreamResponseModelBenchmarkDelta},
}
for _, tt := range tests {
b.Run(tt.name, func(b *testing.B) {
b.Run("legacy", func(b *testing.B) {
b.ReportAllocs()
for b.Loop() {
upstreamResponseModelBenchmarkSink = benchmarkLegacyOpenAIModel(tt.payload)
}
})
b.Run("optimized", func(b *testing.B) {
b.ReportAllocs()
for b.Loop() {
upstreamResponseModelBenchmarkSink = firstValidTrimmedGJSONString(tt.payload, "response.model", "model")
}
})
})
}
}
func BenchmarkUpstreamResponseModelAntigravityWrapper(b *testing.B) {
b.Run("legacy", func(b *testing.B) {
b.ReportAllocs()
for b.Loop() {
upstreamResponseModelBenchmarkSink = benchmarkLegacyWrappedGeminiModel(upstreamResponseModelBenchmarkWrapper)
}
})
b.Run("optimized", func(b *testing.B) {
b.ReportAllocs()
for b.Loop() {
upstreamResponseModelBenchmarkSink = firstValidTrimmedGJSONString(
upstreamResponseModelBenchmarkWrapper,
"modelVersion",
"response.modelVersion",
"response.response.modelVersion",
)
}
})
}
func benchmarkLegacyOpenAIModel(payload []byte) string {
if len(payload) == 0 || !gjson.ValidBytes(payload) {
return ""
}
return benchmarkFirstTrimmedGJSONModel(
gjson.GetBytes(payload, "response.model"),
gjson.GetBytes(payload, "model"),
)
}
func benchmarkLegacyWrappedGeminiModel(payload []byte) string {
if inner := gjson.GetBytes(payload, "response"); inner.Exists() {
payload = []byte(inner.Raw)
}
if len(payload) == 0 || !gjson.ValidBytes(payload) {
return ""
}
return benchmarkFirstTrimmedGJSONModel(
gjson.GetBytes(payload, "modelVersion"),
gjson.GetBytes(payload, "response.modelVersion"),
)
}
func benchmarkFirstTrimmedGJSONModel(values ...gjson.Result) string {
for _, value := range values {
if !value.Exists() || value.Type != gjson.String {
continue
}
if model := strings.TrimSpace(value.String()); model != "" {
return model
}
}
return ""
}