Merge pull request #6379 from baryon/contrib/codex-priority-service-tier

fix(codex): advertise Fast service tier in generated catalogs

backend/internal/service/openai_codex_models_service.go

@@ -324,6 +324,12 @@ type configuredCodexTruncationPolicy struct {
 	Limit int64  `json:"limit"`
 }
 
+type configuredCodexServiceTier struct {
+	ID          string `json:"id"`
+	Name        string `json:"name"`
+	Description string `json:"description"`
+}
+
 type configuredCodexModelMessages struct {
 	InstructionsTemplate  string `json:"instructions_template"`
 	InstructionsVariables any    `json:"instructions_variables"`
@@ -351,7 +357,7 @@ type configuredCodexModelDescriptor struct {
 	SupportedInAPI                    bool                            `json:"supported_in_api"`
 	Priority                          int                             `json:"priority"`
 	AdditionalSpeedTiers              []string                        `json:"additional_speed_tiers"`
-	ServiceTiers                      []any                           `json:"service_tiers"`
+	ServiceTiers                      []configuredCodexServiceTier    `json:"service_tiers"`
 	DefaultServiceTier                any                             `json:"default_service_tier"`
 	AvailabilityNUX                   any                             `json:"availability_nux"`
 	Upgrade                           any                             `json:"upgrade"`
@@ -407,7 +413,7 @@ func newConfiguredCodexModelDescriptor(modelID string) configuredCodexModelDescr
 		SupportedInAPI:                    true,
 		Priority:                          configuredCodexModelPriority,
 		AdditionalSpeedTiers:              []string{},
-		ServiceTiers:                      []any{},
+		ServiceTiers:                      []configuredCodexServiceTier{},
 		ModelMessages:                     configuredCodexModelMessages{InstructionsTemplate: openai.CodexBaseInstructionsForModel(modelID)},
 		SupportsReasoningSummaryParameter: true,
 		DefaultReasoningSummary:           "auto",
@@ -463,6 +469,15 @@ func newConfiguredCodexModelDescriptor(modelID string) configuredCodexModelDescr
 		descriptor.DisplayName = openaiCodexDisplayName(modelID)
 		descriptor.Description = "OpenAI GPT coding model routed through Sub2API."
 		descriptor.SupportsParallelToolCalls = true
+		if configuredCodexSupportsPriorityServiceTier(modelID) {
+			descriptor.ServiceTiers = []configuredCodexServiceTier{
+				{
+					ID:          "priority",
+					Name:        "Fast",
+					Description: "Priority processing for lower latency.",
+				},
+			}
+		}
 		if isOpenAICodexReasoningGPTModel(modelID) {
 			defaultReasoningLevel := "medium"
 			if getNormalizedCodexModel(modelID) == "gpt-5.6-sol" {
@@ -486,6 +501,16 @@ func newConfiguredCodexModelDescriptor(modelID string) configuredCodexModelDescr
 	return descriptor
 }
 
+func configuredCodexSupportsPriorityServiceTier(modelID string) bool {
+	normalized := canonicalizeOpenAIModelAliasSpelling(modelID)
+	for _, family := range []string{"gpt-5.4", "gpt-5.5", "gpt-5.6"} {
+		if normalized == family || strings.HasPrefix(normalized, family+"-") {
+			return true
+		}
+	}
+	return false
+}
+
 func configuredCodexGrokReasoningLevels(modelID string) []configuredCodexReasoningLevel {
 	levels := []configuredCodexReasoningLevel{
 		{Effort: "low", Description: "Fast responses with lighter reasoning"},

backend/internal/service/openai_codex_models_service_test.go

@@ -205,7 +205,7 @@ func requireCompleteConfiguredCodexModel(t *testing.T, model map[string]any, slu
 	require.Equal(t, true, model["supported_in_api"])
 	require.NotNil(t, model["priority"])
 	require.Equal(t, []any{}, model["additional_speed_tiers"])
-	require.Equal(t, []any{}, model["service_tiers"])
+	require.IsType(t, []any{}, model["service_tiers"])
 	require.Contains(t, model, "default_service_tier")
 	require.Contains(t, model, "availability_nux")
 	require.Contains(t, model, "upgrade")
@@ -427,6 +427,51 @@ func TestBuildCodexModelsManifestKeepsKnownReasoningChoices(t *testing.T) {
 	require.NotEqual(t, "none", firstLevel["effort"])
 }
 
+// Scenario: 支持 Fast 的 GPT 型号在目录中声明 priority service tier。
+func TestBuildCodexModelsManifestAdvertisesPriorityServiceTierForFastGPTModels(t *testing.T) {
+	t.Parallel()
+
+	body, err := BuildCodexModelsManifest([]string{
+		"gpt-5.4-mini",
+		"gpt-5.5",
+		"gpt-5.6-sol",
+	})
+	require.NoError(t, err)
+	models := decodeCodexManifestModels(t, body)
+	require.Len(t, models, 3)
+
+	for _, model := range models {
+		require.Equal(t, []any{
+			map[string]any{
+				"id":          "priority",
+				"name":        "Fast",
+				"description": "Priority processing for lower latency.",
+			},
+		}, model["service_tiers"])
+		require.Nil(t, model["default_service_tier"])
+	}
+}
+
+// Scenario: 未明确支持 Fast 的型号不推测 service tier。
+func TestBuildCodexModelsManifestLeavesServiceTiersEmptyForOtherModels(t *testing.T) {
+	t.Parallel()
+
+	body, err := BuildCodexModelsManifest([]string{
+		"gpt-4o",
+		"claude-opus-4-6",
+		"deepseek-v4-pro",
+		"company-coding-model",
+	})
+	require.NoError(t, err)
+	models := decodeCodexManifestModels(t, body)
+	require.Len(t, models, 4)
+
+	for _, model := range models {
+		require.Equal(t, []any{}, model["service_tiers"])
+		require.Nil(t, model["default_service_tier"])
+	}
+}
+
 // Scenario: 专用图片生成模型不进入 Codex 主模型目录。
 func TestBuildCodexModelsManifestOmitsDedicatedImageModels(t *testing.T) {
 	t.Parallel()
@@ -1895,7 +1940,7 @@ func TestCompleteAPIKeyCodexModelsManifestForClientPreservesProviderMetadata(t *
 
 	svc := &OpenAIGatewayService{}
 	manifest := &CodexModelsManifest{
-		Body: []byte(`{"models":[{"slug":"grok-4.6","description":"Provider supplied","model_messages":{"auto_review":{"enabled":true}},"truncation_policy":{"mode":"tokens"},"unknown":{"kept":true}}],"metadata":{"source":"upstream"}}`),
+		Body: []byte(`{"models":[{"slug":"grok-4.6","description":"Provider supplied","service_tiers":[{"id":"provider-priority","name":"Provider Fast","description":"Provider supplied tier."}],"model_messages":{"auto_review":{"enabled":true}},"truncation_policy":{"mode":"tokens"},"unknown":{"kept":true}}],"metadata":{"source":"upstream"}}`),
 	}
 	account := newCodexModelsAPIKeyTestAccount("https://upstream.example/v1")
 
@@ -1905,6 +1950,13 @@ func TestCompleteAPIKeyCodexModelsManifestForClientPreservesProviderMetadata(t *
 	requireCompleteConfiguredCodexModel(t, models[0], "grok-4.6")
 	require.Equal(t, "Provider supplied", models[0]["description"])
 	require.Equal(t, map[string]any{"kept": true}, models[0]["unknown"])
+	require.Equal(t, []any{
+		map[string]any{
+			"id":          "provider-priority",
+			"name":        "Provider Fast",
+			"description": "Provider supplied tier.",
+		},
+	}, models[0]["service_tiers"])
 	require.Equal(t, []any{"text"}, models[0]["input_modalities"])
 	require.Equal(t, []string{"low", "medium", "high", "xhigh"}, effortsFromManifestModel(t, models[0]))
 	modelMessages, ok := models[0]["model_messages"].(map[string]any)