fix(openai): preserve explicit GPT-5.6 cache write prices

backend/internal/service/billing_service.go

@@ -97,6 +97,7 @@ type ModelPricing struct {
 	OutputPricePerTokenPriority        float64 // priority service tier 下每token输出价格 (USD)
 	CacheCreationPricePerToken         float64 // 缓存创建每token价格 (USD)
 	CacheCreationPricePerTokenPriority float64 // priority service tier 下缓存创建每token价格 (USD)
+	CacheCreationPriceExplicit         bool    // 是否由渠道/区间定价显式设定（为 true 时即使 == 0 也不回退）
 	CacheReadPricePerToken             float64 // 缓存读取每token价格 (USD)
 	CacheReadPricePerTokenPriority     float64 // priority service tier 下缓存读取每token价格 (USD)
 	CacheCreation5mPrice               float64 // 5分钟缓存创建每token价格 (USD)
@@ -825,6 +826,7 @@ func (s *BillingService) GetModelPricingWithChannel(model string, channelPricing
 	if channelPricing.CacheWritePrice != nil {
 		pricing.CacheCreationPricePerToken = *channelPricing.CacheWritePrice
 		pricing.CacheCreationPricePerTokenPriority = *channelPricing.CacheWritePrice
+		pricing.CacheCreationPriceExplicit = true
 		pricing.CacheCreation5mPrice = *channelPricing.CacheWritePrice
 		pricing.CacheCreation1hPrice = *channelPricing.CacheWritePrice
 	}
@@ -1100,7 +1102,7 @@ func (s *BillingService) applyModelSpecificPricingPolicy(model string, pricing *
 	}
 	needsLongContextPolicy := usesLegacyLongContextPricing &&
 		(pricing.LongContextInputThreshold <= 0 || pricing.LongContextInputMultiplier <= 0 || pricing.LongContextOutputMultiplier <= 0)
-	needsCacheCreationPolicy := isGPT56 && (pricing.CacheCreationPricePerToken <= 0 ||
+	needsCacheCreationPolicy := isGPT56 && !pricing.CacheCreationPriceExplicit && (pricing.CacheCreationPricePerToken <= 0 ||
 		(pricing.InputPricePerTokenPriority > 0 && pricing.CacheCreationPricePerTokenPriority <= 0))
 	if !needsLongContextPolicy && !needsCacheCreationPolicy {
 		return pricing

backend/internal/service/model_pricing_resolver.go

@@ -184,6 +184,7 @@ func (r *ModelPricingResolver) applyTokenOverrides(chPricing *ChannelModelPricin
 	if chPricing.CacheWritePrice != nil {
 		resolved.BasePricing.CacheCreationPricePerToken = *chPricing.CacheWritePrice
 		resolved.BasePricing.CacheCreationPricePerTokenPriority = *chPricing.CacheWritePrice
+		resolved.BasePricing.CacheCreationPriceExplicit = true
 		resolved.BasePricing.CacheCreation5mPrice = *chPricing.CacheWritePrice
 		resolved.BasePricing.CacheCreation1hPrice = *chPricing.CacheWritePrice
 	}
@@ -253,6 +254,7 @@ func intervalToModelPricing(iv *PricingInterval, supportsCacheBreakdown bool, ch
 	if iv.CacheWritePrice != nil {
 		pricing.CacheCreationPricePerToken = *iv.CacheWritePrice
 		pricing.CacheCreationPricePerTokenPriority = *iv.CacheWritePrice
+		pricing.CacheCreationPriceExplicit = true
 		pricing.CacheCreation5mPrice = *iv.CacheWritePrice
 		pricing.CacheCreation1hPrice = *iv.CacheWritePrice
 	}

backend/internal/service/model_pricing_resolver_test.go

@@ -114,6 +114,52 @@ func TestGetIntervalPricing_NoMatch_FallsBackToBase(t *testing.T) {
 	require.Equal(t, basePricing, result)
 }
 
+func TestGPT56ExplicitZeroCacheWritePriceIsPreserved(t *testing.T) {
+	bs := &BillingService{}
+	resolver := NewModelPricingResolver(nil, bs)
+	zero := 0.0
+
+	t.Run("flat channel price", func(t *testing.T) {
+		resolved := &ResolvedPricing{
+			Mode: BillingModeToken,
+			BasePricing: &ModelPricing{
+				InputPricePerToken:  5e-6,
+				OutputPricePerToken: 30e-6,
+			},
+		}
+		resolver.applyTokenOverrides(&ChannelModelPricing{CacheWritePrice: &zero}, resolved)
+
+		require.True(t, resolved.BasePricing.CacheCreationPriceExplicit)
+		cost, err := bs.CalculateCostUnified(CostInput{
+			Model:          "gpt-5.6-sol",
+			Tokens:         UsageTokens{CacheCreationTokens: 100},
+			RateMultiplier: 1,
+			Resolver:       resolver,
+			Resolved:       resolved,
+		})
+		require.NoError(t, err)
+		require.Zero(t, cost.CacheCreationCost)
+	})
+
+	t.Run("interval price", func(t *testing.T) {
+		pricing := intervalToModelPricing(&PricingInterval{CacheWritePrice: &zero}, false, nil)
+		require.True(t, pricing.CacheCreationPriceExplicit)
+
+		cost, err := bs.CalculateCostUnified(CostInput{
+			Model:          "gpt-5.6-sol",
+			Tokens:         UsageTokens{CacheCreationTokens: 100},
+			RateMultiplier: 1,
+			Resolver:       resolver,
+			Resolved: &ResolvedPricing{
+				Mode:        BillingModeToken,
+				BasePricing: pricing,
+			},
+		})
+		require.NoError(t, err)
+		require.Zero(t, cost.CacheCreationCost)
+	})
+}
+
 func TestGetRequestTierPrice(t *testing.T) {
 	bs := newTestBillingServiceForResolver()
 	r := NewModelPricingResolver(&ChannelService{}, bs)