Fixed: #478

feat(antigravity): add support for inline image data in client responses

internal/api/modules/amp/routes.go

@@ -156,6 +156,7 @@ func (m *AmpModule) registerManagementRoutes(engine *gin.Engine, baseHandler *ha
 	rootMiddleware := []gin.HandlerFunc{m.managementAvailabilityMiddleware(), noCORSMiddleware(), m.localhostOnlyMiddleware()}
 	engine.GET("/threads/*path", append(rootMiddleware, proxyHandler)...)
 	engine.GET("/threads.rss", append(rootMiddleware, proxyHandler)...)
+	engine.GET("/news.rss", append(rootMiddleware, proxyHandler)...)
 
 	// Root-level auth routes for CLI login flow
 	// Amp uses multiple auth routes: /auth/cli-login, /auth/callback, /auth/sign-in, /auth/logout

internal/registry/model_definitions.go

@@ -693,8 +693,8 @@ func GetOpenAIModels() []*ModelInfo {
 			OwnedBy:             "openai",
 			Type:                "openai",
 			Version:             "gpt-5.1-2025-11-12",
-			DisplayName:         "GPT 5 Low",
-			Description:         "Stable version of GPT 5, The best model for coding and agentic tasks across domains.",
+			DisplayName:         "GPT 5.1 Nothink",
+			Description:         "Stable version of GPT 5.1, The best model for coding and agentic tasks across domains.",
 			ContextLength:       400000,
 			MaxCompletionTokens: 128000,
 			SupportedParameters: []string{"tools"},
@@ -719,8 +719,8 @@ func GetOpenAIModels() []*ModelInfo {
 			OwnedBy:             "openai",
 			Type:                "openai",
 			Version:             "gpt-5.1-2025-11-12",
-			DisplayName:         "GPT 5 Medium",
-			Description:         "Stable version of GPT 5, The best model for coding and agentic tasks across domains.",
+			DisplayName:         "GPT 5.1 Medium",
+			Description:         "Stable version of GPT 5.1, The best model for coding and agentic tasks across domains.",
 			ContextLength:       400000,
 			MaxCompletionTokens: 128000,
 			SupportedParameters: []string{"tools"},
@@ -732,8 +732,8 @@ func GetOpenAIModels() []*ModelInfo {
 			OwnedBy:             "openai",
 			Type:                "openai",
 			Version:             "gpt-5.1-2025-11-12",
-			DisplayName:         "GPT 5 High",
-			Description:         "Stable version of GPT 5, The best model for coding and agentic tasks across domains.",
+			DisplayName:         "GPT 5.1 High",
+			Description:         "Stable version of GPT 5.1, The best model for coding and agentic tasks across domains.",
 			ContextLength:       400000,
 			MaxCompletionTokens: 128000,
 			SupportedParameters: []string{"tools"},
@@ -745,8 +745,8 @@ func GetOpenAIModels() []*ModelInfo {
 			OwnedBy:             "openai",
 			Type:                "openai",
 			Version:             "gpt-5.1-2025-11-12",
-			DisplayName:         "GPT 5 Codex",
-			Description:         "Stable version of GPT 5 Codex, The best model for coding and agentic tasks across domains.",
+			DisplayName:         "GPT 5.1 Codex",
+			Description:         "Stable version of GPT 5.1 Codex, The best model for coding and agentic tasks across domains.",
 			ContextLength:       400000,
 			MaxCompletionTokens: 128000,
 			SupportedParameters: []string{"tools"},
@@ -758,8 +758,8 @@ func GetOpenAIModels() []*ModelInfo {
 			OwnedBy:             "openai",
 			Type:                "openai",
 			Version:             "gpt-5.1-2025-11-12",
-			DisplayName:         "GPT 5 Codex Low",
-			Description:         "Stable version of GPT 5 Codex, The best model for coding and agentic tasks across domains.",
+			DisplayName:         "GPT 5.1 Codex Low",
+			Description:         "Stable version of GPT 5.1 Codex, The best model for coding and agentic tasks across domains.",
 			ContextLength:       400000,
 			MaxCompletionTokens: 128000,
 			SupportedParameters: []string{"tools"},
@@ -771,8 +771,8 @@ func GetOpenAIModels() []*ModelInfo {
 			OwnedBy:             "openai",
 			Type:                "openai",
 			Version:             "gpt-5.1-2025-11-12",
-			DisplayName:         "GPT 5 Codex Medium",
-			Description:         "Stable version of GPT 5 Codex, The best model for coding and agentic tasks across domains.",
+			DisplayName:         "GPT 5.1 Codex Medium",
+			Description:         "Stable version of GPT 5.1 Codex, The best model for coding and agentic tasks across domains.",
 			ContextLength:       400000,
 			MaxCompletionTokens: 128000,
 			SupportedParameters: []string{"tools"},
@@ -784,8 +784,8 @@ func GetOpenAIModels() []*ModelInfo {
 			OwnedBy:             "openai",
 			Type:                "openai",
 			Version:             "gpt-5.1-2025-11-12",
-			DisplayName:         "GPT 5 Codex High",
-			Description:         "Stable version of GPT 5 Codex, The best model for coding and agentic tasks across domains.",
+			DisplayName:         "GPT 5.1 Codex High",
+			Description:         "Stable version of GPT 5.1 Codex, The best model for coding and agentic tasks across domains.",
 			ContextLength:       400000,
 			MaxCompletionTokens: 128000,
 			SupportedParameters: []string{"tools"},
@@ -797,8 +797,8 @@ func GetOpenAIModels() []*ModelInfo {
 			OwnedBy:             "openai",
 			Type:                "openai",
 			Version:             "gpt-5.1-2025-11-12",
-			DisplayName:         "GPT 5 Codex Mini",
-			Description:         "Stable version of GPT 5 Codex Mini: cheaper, faster, but less capable version of GPT 5 Codex.",
+			DisplayName:         "GPT 5.1 Codex Mini",
+			Description:         "Stable version of GPT 5.1 Codex Mini: cheaper, faster, but less capable version of GPT 5.1 Codex.",
 			ContextLength:       400000,
 			MaxCompletionTokens: 128000,
 			SupportedParameters: []string{"tools"},
@@ -810,8 +810,8 @@ func GetOpenAIModels() []*ModelInfo {
 			OwnedBy:             "openai",
 			Type:                "openai",
 			Version:             "gpt-5.1-2025-11-12",
-			DisplayName:         "GPT 5 Codex Mini Medium",
-			Description:         "Stable version of GPT 5 Codex Mini: cheaper, faster, but less capable version of GPT 5 Codex.",
+			DisplayName:         "GPT 5.1 Codex Mini Medium",
+			Description:         "Stable version of GPT 5.1 Codex Mini: cheaper, faster, but less capable version of GPT 5.1 Codex.",
 			ContextLength:       400000,
 			MaxCompletionTokens: 128000,
 			SupportedParameters: []string{"tools"},
@@ -823,8 +823,8 @@ func GetOpenAIModels() []*ModelInfo {
 			OwnedBy:             "openai",
 			Type:                "openai",
 			Version:             "gpt-5.1-2025-11-12",
-			DisplayName:         "GPT 5 Codex Mini High",
-			Description:         "Stable version of GPT 5 Codex Mini: cheaper, faster, but less capable version of GPT 5 Codex.",
+			DisplayName:         "GPT 5.1 Codex Mini High",
+			Description:         "Stable version of GPT 5.1 Codex Mini: cheaper, faster, but less capable version of GPT 5.1 Codex.",
 			ContextLength:       400000,
 			MaxCompletionTokens: 128000,
 			SupportedParameters: []string{"tools"},
@@ -837,8 +837,8 @@ func GetOpenAIModels() []*ModelInfo {
 			OwnedBy:             "openai",
 			Type:                "openai",
 			Version:             "gpt-5.1-max",
-			DisplayName:         "GPT 5 Codex Max",
-			Description:         "Stable version of GPT 5 Codex Max",
+			DisplayName:         "GPT 5.1 Codex Max",
+			Description:         "Stable version of GPT 5.1 Codex Max",
 			ContextLength:       400000,
 			MaxCompletionTokens: 128000,
 			SupportedParameters: []string{"tools"},
@@ -850,8 +850,8 @@ func GetOpenAIModels() []*ModelInfo {
 			OwnedBy:             "openai",
 			Type:                "openai",
 			Version:             "gpt-5.1-max",
-			DisplayName:         "GPT 5 Codex Max Low",
-			Description:         "Stable version of GPT 5 Codex Max Low",
+			DisplayName:         "GPT 5.1 Codex Max Low",
+			Description:         "Stable version of GPT 5.1 Codex Max Low",
 			ContextLength:       400000,
 			MaxCompletionTokens: 128000,
 			SupportedParameters: []string{"tools"},
@@ -863,8 +863,8 @@ func GetOpenAIModels() []*ModelInfo {
 			OwnedBy:             "openai",
 			Type:                "openai",
 			Version:             "gpt-5.1-max",
-			DisplayName:         "GPT 5 Codex Max Medium",
-			Description:         "Stable version of GPT 5 Codex Max Medium",
+			DisplayName:         "GPT 5.1 Codex Max Medium",
+			Description:         "Stable version of GPT 5.1 Codex Max Medium",
 			ContextLength:       400000,
 			MaxCompletionTokens: 128000,
 			SupportedParameters: []string{"tools"},
@@ -876,8 +876,8 @@ func GetOpenAIModels() []*ModelInfo {
 			OwnedBy:             "openai",
 			Type:                "openai",
 			Version:             "gpt-5.1-max",
-			DisplayName:         "GPT 5 Codex Max High",
-			Description:         "Stable version of GPT 5 Codex Max High",
+			DisplayName:         "GPT 5.1 Codex Max High",
+			Description:         "Stable version of GPT 5.1 Codex Max High",
 			ContextLength:       400000,
 			MaxCompletionTokens: 128000,
 			SupportedParameters: []string{"tools"},
@@ -889,8 +889,8 @@ func GetOpenAIModels() []*ModelInfo {
 			OwnedBy:             "openai",
 			Type:                "openai",
 			Version:             "gpt-5.1-max",
-			DisplayName:         "GPT 5 Codex Max XHigh",
-			Description:         "Stable version of GPT 5 Codex Max XHigh",
+			DisplayName:         "GPT 5.1 Codex Max XHigh",
+			Description:         "Stable version of GPT 5.1 Codex Max XHigh",
 			ContextLength:       400000,
 			MaxCompletionTokens: 128000,
 			SupportedParameters: []string{"tools"},
@@ -991,16 +991,19 @@ func GetIFlowModels() []*ModelInfo {
 type AntigravityModelConfig struct {
 	Thinking            *ThinkingSupport
 	MaxCompletionTokens int
+	Name                string
 }
 
 // GetAntigravityModelConfig returns static configuration for antigravity models.
 // Keys use the ALIASED model names (after modelName2Alias conversion) for direct lookup.
 func GetAntigravityModelConfig() map[string]*AntigravityModelConfig {
 	return map[string]*AntigravityModelConfig{
-		"gemini-2.5-flash":                  {Thinking: &ThinkingSupport{Min: 0, Max: 24576, ZeroAllowed: true, DynamicAllowed: true}},
-		"gemini-2.5-flash-lite":             {Thinking: &ThinkingSupport{Min: 0, Max: 24576, ZeroAllowed: true, DynamicAllowed: true}},
-		"gemini-3-pro-preview":              {Thinking: &ThinkingSupport{Min: 128, Max: 32768, ZeroAllowed: false, DynamicAllowed: true}},
-		"gemini-claude-sonnet-4-5-thinking": {Thinking: &ThinkingSupport{Min: 1024, Max: 200000, ZeroAllowed: false, DynamicAllowed: true}, MaxCompletionTokens: 64000},
-		"gemini-claude-opus-4-5-thinking":   {Thinking: &ThinkingSupport{Min: 1024, Max: 200000, ZeroAllowed: false, DynamicAllowed: true}, MaxCompletionTokens: 64000},
+		"gemini-2.5-flash":                        {Thinking: &ThinkingSupport{Min: 0, Max: 24576, ZeroAllowed: true, DynamicAllowed: true}, Name: "models/gemini-2.5-flash"},
+		"gemini-2.5-flash-lite":                   {Thinking: &ThinkingSupport{Min: 0, Max: 24576, ZeroAllowed: true, DynamicAllowed: true}, Name: "models/gemini-2.5-flash-lite"},
+		"gemini-2.5-computer-use-preview-10-2025": {Name: "models/gemini-2.5-computer-use-preview-10-2025"},
+		"gemini-3-pro-preview":                    {Thinking: &ThinkingSupport{Min: 128, Max: 32768, ZeroAllowed: false, DynamicAllowed: true}, Name: "models/gemini-3-pro-preview"},
+		"gemini-3-pro-image-preview":              {Thinking: &ThinkingSupport{Min: 128, Max: 32768, ZeroAllowed: false, DynamicAllowed: true}, Name: "models/gemini-3-pro-image-preview"},
+		"gemini-claude-sonnet-4-5-thinking":       {Thinking: &ThinkingSupport{Min: 1024, Max: 200000, ZeroAllowed: false, DynamicAllowed: true}, MaxCompletionTokens: 64000},
+		"gemini-claude-opus-4-5-thinking":         {Thinking: &ThinkingSupport{Min: 1024, Max: 200000, ZeroAllowed: false, DynamicAllowed: true}, MaxCompletionTokens: 64000},
 	}
 }

internal/runtime/executor/aistudio_executor.go

@@ -309,11 +309,9 @@ func (e *AIStudioExecutor) translateRequest(req cliproxyexecutor.Request, opts c
 	to := sdktranslator.FromString("gemini")
 	payload := sdktranslator.TranslateRequest(from, to, req.Model, bytes.Clone(req.Payload), stream)
 	payload = applyThinkingMetadata(payload, req.Metadata, req.Model)
+	payload = util.ApplyDefaultThinkingIfNeeded(req.Model, payload)
 	payload = util.ConvertThinkingLevelToBudget(payload)
-	if budget := gjson.GetBytes(payload, "generationConfig.thinkingConfig.thinkingBudget"); budget.Exists() {
-		normalized := util.NormalizeThinkingBudget(req.Model, int(budget.Int()))
-		payload, _ = sjson.SetBytes(payload, "generationConfig.thinkingConfig.thinkingBudget", normalized)
-	}
+	payload = util.NormalizeGeminiThinkingBudget(req.Model, payload)
 	payload = util.StripThinkingConfigIfUnsupported(req.Model, payload)
 	payload = fixGeminiImageAspectRatio(req.Model, payload)
 	payload = applyPayloadConfig(e.cfg, req.Model, payload)

internal/runtime/executor/antigravity_executor.go

@@ -27,18 +27,18 @@ import (
 )
 
 const (
-	antigravityBaseURLDaily        = "https://daily-cloudcode-pa.sandbox.googleapis.com"
-	antigravityBaseURLAutopush     = "https://autopush-cloudcode-pa.sandbox.googleapis.com"
-	antigravityBaseURLProd         = "https://cloudcode-pa.googleapis.com"
-	antigravityStreamPath          = "/v1internal:streamGenerateContent"
-	antigravityGeneratePath        = "/v1internal:generateContent"
-	antigravityModelsPath          = "/v1internal:fetchAvailableModels"
-	antigravityClientID            = "1071006060591-tmhssin2h21lcre235vtolojh4g403ep.apps.googleusercontent.com"
-	antigravityClientSecret        = "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"
-	defaultAntigravityAgent        = "antigravity/1.11.5 windows/amd64"
-	antigravityAuthType            = "antigravity"
-	refreshSkew                    = 3000 * time.Second
-	streamScannerBuffer        int = 20_971_520
+	antigravityBaseURLDaily = "https://daily-cloudcode-pa.sandbox.googleapis.com"
+	// antigravityBaseURLAutopush     = "https://autopush-cloudcode-pa.sandbox.googleapis.com"
+	antigravityBaseURLProd      = "https://cloudcode-pa.googleapis.com"
+	antigravityStreamPath       = "/v1internal:streamGenerateContent"
+	antigravityGeneratePath     = "/v1internal:generateContent"
+	antigravityModelsPath       = "/v1internal:fetchAvailableModels"
+	antigravityClientID         = "1071006060591-tmhssin2h21lcre235vtolojh4g403ep.apps.googleusercontent.com"
+	antigravityClientSecret     = "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"
+	defaultAntigravityAgent     = "antigravity/1.11.5 windows/amd64"
+	antigravityAuthType         = "antigravity"
+	refreshSkew                 = 3000 * time.Second
+	streamScannerBuffer     int = 20_971_520
 )
 
 var randSource = rand.New(rand.NewSource(time.Now().UnixNano()))
@@ -77,6 +77,7 @@ func (e *AntigravityExecutor) Execute(ctx context.Context, auth *cliproxyauth.Au
 	translated := sdktranslator.TranslateRequest(from, to, req.Model, bytes.Clone(req.Payload), false)
 
 	translated = applyThinkingMetadataCLI(translated, req.Metadata, req.Model)
+	translated = util.ApplyDefaultThinkingIfNeededCLI(req.Model, translated)
 	translated = normalizeAntigravityThinking(req.Model, translated)
 
 	baseURLs := antigravityBaseURLFallbackOrder(auth)
@@ -171,6 +172,7 @@ func (e *AntigravityExecutor) ExecuteStream(ctx context.Context, auth *cliproxya
 	translated := sdktranslator.TranslateRequest(from, to, req.Model, bytes.Clone(req.Payload), true)
 
 	translated = applyThinkingMetadataCLI(translated, req.Metadata, req.Model)
+	translated = util.ApplyDefaultThinkingIfNeededCLI(req.Model, translated)
 	translated = normalizeAntigravityThinking(req.Model, translated)
 
 	baseURLs := antigravityBaseURLFallbackOrder(auth)
@@ -373,9 +375,14 @@ func FetchAntigravityModels(ctx context.Context, auth *cliproxyauth.Auth, cfg *c
 		for originalName := range result.Map() {
 			aliasName := modelName2Alias(originalName)
 			if aliasName != "" {
+				cfg := modelConfig[aliasName]
+				modelName := aliasName
+				if cfg != nil && cfg.Name != "" {
+					modelName = cfg.Name
+				}
 				modelInfo := &registry.ModelInfo{
 					ID:          aliasName,
-					Name:        aliasName,
+					Name:        modelName,
 					Description: aliasName,
 					DisplayName: aliasName,
 					Version:     aliasName,
@@ -385,7 +392,7 @@ func FetchAntigravityModels(ctx context.Context, auth *cliproxyauth.Auth, cfg *c
 					Type:        antigravityAuthType,
 				}
 				// Look up Thinking support from static config using alias name
-				if cfg, ok := modelConfig[aliasName]; ok {
+				if cfg != nil {
 					if cfg.Thinking != nil {
 						modelInfo.Thinking = cfg.Thinking
 					}
@@ -654,7 +661,7 @@ func buildBaseURL(auth *cliproxyauth.Auth) string {
 	if baseURLs := antigravityBaseURLFallbackOrder(auth); len(baseURLs) > 0 {
 		return baseURLs[0]
 	}
-	return antigravityBaseURLAutopush
+	return antigravityBaseURLDaily
 }
 
 func resolveHost(base string) string {
@@ -690,7 +697,7 @@ func antigravityBaseURLFallbackOrder(auth *cliproxyauth.Auth) []string {
 	}
 	return []string{
 		antigravityBaseURLDaily,
-		antigravityBaseURLAutopush,
+		// antigravityBaseURLAutopush,
 		antigravityBaseURLProd,
 	}
 }
@@ -834,9 +841,12 @@ func normalizeAntigravityThinking(model string, payload []byte) []byte {
 		effectiveMax, setDefaultMax := antigravityEffectiveMaxTokens(model, payload)
 		if effectiveMax > 0 && normalized >= effectiveMax {
 			normalized = effectiveMax - 1
-			if normalized < 1 {
-				normalized = 1
-			}
+		}
+		minBudget := antigravityMinThinkingBudget(model)
+		if minBudget > 0 && normalized >= 0 && normalized < minBudget {
+			// Budget is below minimum, remove thinking config entirely
+			payload, _ = sjson.DeleteBytes(payload, "request.generationConfig.thinkingConfig")
+			return payload
 		}
 		if setDefaultMax {
 			if res, errSet := sjson.SetBytes(payload, "request.generationConfig.maxOutputTokens", effectiveMax); errSet == nil {
@@ -864,3 +874,12 @@ func antigravityEffectiveMaxTokens(model string, payload []byte) (max int, fromM
 	}
 	return 0, false
 }
+
+// antigravityMinThinkingBudget returns the minimum thinking budget for a model.
+// Falls back to -1 if no model info is found.
+func antigravityMinThinkingBudget(model string) int {
+	if modelInfo := registry.GetGlobalRegistry().GetModelInfo(model); modelInfo != nil && modelInfo.Thinking != nil {
+		return modelInfo.Thinking.Min
+	}
+	return -1
+}

internal/runtime/executor/gemini_cli_executor.go

@@ -64,6 +64,8 @@ func (e *GeminiCLIExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth
 	to := sdktranslator.FromString("gemini-cli")
 	basePayload := sdktranslator.TranslateRequest(from, to, req.Model, bytes.Clone(req.Payload), false)
 	basePayload = applyThinkingMetadataCLI(basePayload, req.Metadata, req.Model)
+	basePayload = util.ApplyDefaultThinkingIfNeededCLI(req.Model, basePayload)
+	basePayload = util.NormalizeGeminiCLIThinkingBudget(req.Model, basePayload)
 	basePayload = util.StripThinkingConfigIfUnsupported(req.Model, basePayload)
 	basePayload = fixGeminiCLIImageAspectRatio(req.Model, basePayload)
 	basePayload = applyPayloadConfigWithRoot(e.cfg, req.Model, "gemini", "request", basePayload)
@@ -199,6 +201,8 @@ func (e *GeminiCLIExecutor) ExecuteStream(ctx context.Context, auth *cliproxyaut
 	to := sdktranslator.FromString("gemini-cli")
 	basePayload := sdktranslator.TranslateRequest(from, to, req.Model, bytes.Clone(req.Payload), true)
 	basePayload = applyThinkingMetadataCLI(basePayload, req.Metadata, req.Model)
+	basePayload = util.ApplyDefaultThinkingIfNeededCLI(req.Model, basePayload)
+	basePayload = util.NormalizeGeminiCLIThinkingBudget(req.Model, basePayload)
 	basePayload = util.StripThinkingConfigIfUnsupported(req.Model, basePayload)
 	basePayload = fixGeminiCLIImageAspectRatio(req.Model, basePayload)
 	basePayload = applyPayloadConfigWithRoot(e.cfg, req.Model, "gemini", "request", basePayload)

internal/runtime/executor/gemini_executor.go

@@ -80,6 +80,8 @@ func (e *GeminiExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, r
 	to := sdktranslator.FromString("gemini")
 	body := sdktranslator.TranslateRequest(from, to, req.Model, bytes.Clone(req.Payload), false)
 	body = applyThinkingMetadata(body, req.Metadata, req.Model)
+	body = util.ApplyDefaultThinkingIfNeeded(req.Model, body)
+	body = util.NormalizeGeminiThinkingBudget(req.Model, body)
 	body = util.StripThinkingConfigIfUnsupported(req.Model, body)
 	body = fixGeminiImageAspectRatio(req.Model, body)
 	body = applyPayloadConfig(e.cfg, req.Model, body)
@@ -169,6 +171,8 @@ func (e *GeminiExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.A
 	to := sdktranslator.FromString("gemini")
 	body := sdktranslator.TranslateRequest(from, to, req.Model, bytes.Clone(req.Payload), true)
 	body = applyThinkingMetadata(body, req.Metadata, req.Model)
+	body = util.ApplyDefaultThinkingIfNeeded(req.Model, body)
+	body = util.NormalizeGeminiThinkingBudget(req.Model, body)
 	body = util.StripThinkingConfigIfUnsupported(req.Model, body)
 	body = fixGeminiImageAspectRatio(req.Model, body)
 	body = applyPayloadConfig(e.cfg, req.Model, body)

internal/runtime/executor/gemini_vertex_executor.go

@@ -296,6 +296,8 @@ func (e *GeminiVertexExecutor) executeWithServiceAccount(ctx context.Context, au
 		}
 		body = util.ApplyGeminiThinkingConfig(body, budgetOverride, includeOverride)
 	}
+	body = util.ApplyDefaultThinkingIfNeeded(req.Model, body)
+	body = util.NormalizeGeminiThinkingBudget(req.Model, body)
 	body = util.StripThinkingConfigIfUnsupported(req.Model, body)
 	body = fixGeminiImageAspectRatio(req.Model, body)
 	body = applyPayloadConfig(e.cfg, req.Model, body)
@@ -391,6 +393,8 @@ func (e *GeminiVertexExecutor) executeWithAPIKey(ctx context.Context, auth *clip
 		}
 		body = util.ApplyGeminiThinkingConfig(body, budgetOverride, includeOverride)
 	}
+	body = util.ApplyDefaultThinkingIfNeeded(req.Model, body)
+	body = util.NormalizeGeminiThinkingBudget(req.Model, body)
 	body = util.StripThinkingConfigIfUnsupported(req.Model, body)
 	body = fixGeminiImageAspectRatio(req.Model, body)
 	body = applyPayloadConfig(e.cfg, req.Model, body)
@@ -487,6 +491,8 @@ func (e *GeminiVertexExecutor) executeStreamWithServiceAccount(ctx context.Conte
 		}
 		body = util.ApplyGeminiThinkingConfig(body, budgetOverride, includeOverride)
 	}
+	body = util.ApplyDefaultThinkingIfNeeded(req.Model, body)
+	body = util.NormalizeGeminiThinkingBudget(req.Model, body)
 	body = util.StripThinkingConfigIfUnsupported(req.Model, body)
 	body = fixGeminiImageAspectRatio(req.Model, body)
 	body = applyPayloadConfig(e.cfg, req.Model, body)
@@ -599,6 +605,8 @@ func (e *GeminiVertexExecutor) executeStreamWithAPIKey(ctx context.Context, auth
 		}
 		body = util.ApplyGeminiThinkingConfig(body, budgetOverride, includeOverride)
 	}
+	body = util.ApplyDefaultThinkingIfNeeded(req.Model, body)
+	body = util.NormalizeGeminiThinkingBudget(req.Model, body)
 	body = util.StripThinkingConfigIfUnsupported(req.Model, body)
 	body = fixGeminiImageAspectRatio(req.Model, body)
 	body = applyPayloadConfig(e.cfg, req.Model, body)

internal/runtime/executor/logging_helpers.go

@@ -157,7 +157,7 @@ func appendAPIResponseChunk(ctx context.Context, cfg *config.Config, chunk []byt
 	if ginCtx == nil {
 		return
 	}
-	_, attempt := ensureAttempt(ginCtx)
+	attempts, attempt := ensureAttempt(ginCtx)
 	ensureResponseIntro(attempt)
 
 	if !attempt.headersWritten {
@@ -175,6 +175,8 @@ func appendAPIResponseChunk(ctx context.Context, cfg *config.Config, chunk []byt
 	}
 	attempt.response.WriteString(string(data))
 	attempt.bodyHasContent = true
+
+	updateAggregatedResponse(ginCtx, attempts)
 }
 
 func ginContextFrom(ctx context.Context) *gin.Context {

internal/translator/antigravity/claude/antigravity_claude_request.go

@@ -123,6 +123,15 @@ func ConvertClaudeRequestToAntigravity(modelName string, inputRawJSON []byte, _
 							functionResponse := client.FunctionResponse{ID: toolCallID, Name: funcName, Response: map[string]interface{}{"result": responseData}}
 							clientContent.Parts = append(clientContent.Parts, client.Part{FunctionResponse: &functionResponse})
 						}
+					} else if contentTypeResult.Type == gjson.String && contentTypeResult.String() == "image" {
+						sourceResult := contentResult.Get("source")
+						if sourceResult.Get("type").String() == "base64" {
+							inlineData := &client.InlineData{
+								MimeType: sourceResult.Get("media_type").String(),
+								Data:     sourceResult.Get("data").String(),
+							}
+							clientContent.Parts = append(clientContent.Parts, client.Part{InlineData: inlineData})
+						}
 					}
 				}
 				contents = append(contents, clientContent)
@@ -180,7 +189,6 @@ func ConvertClaudeRequestToAntigravity(modelName string, inputRawJSON []byte, _
 		if t.Get("type").String() == "enabled" {
 			if b := t.Get("budget_tokens"); b.Exists() && b.Type == gjson.Number {
 				budget := int(b.Int())
-				budget = util.NormalizeThinkingBudget(modelName, budget)
 				out, _ = sjson.Set(out, "request.generationConfig.thinkingConfig.thinkingBudget", budget)
 				out, _ = sjson.Set(out, "request.generationConfig.thinkingConfig.include_thoughts", true)
 			}

internal/translator/antigravity/claude/antigravity_claude_response.go

@@ -12,6 +12,7 @@ import (
 	"encoding/json"
 	"fmt"
 	"strings"
+	"sync/atomic"
 	"time"
 
 	"github.com/tidwall/gjson"
@@ -36,6 +37,9 @@ type Params struct {
 	HasToolUse           bool   // Indicates if tool use was observed in the stream
 }
 
+// toolUseIDCounter provides a process-wide unique counter for tool use identifiers.
+var toolUseIDCounter uint64
+
 // ConvertAntigravityResponseToClaude performs sophisticated streaming response format conversion.
 // This function implements a complex state machine that translates backend client responses
 // into Claude Code-compatible Server-Sent Events (SSE) format. It manages different response types
@@ -216,7 +220,7 @@ func ConvertAntigravityResponseToClaude(_ context.Context, _ string, originalReq
 
 				// Create the tool use block with unique ID and function details
 				data := fmt.Sprintf(`{"type":"content_block_start","index":%d,"content_block":{"type":"tool_use","id":"","name":"","input":{}}}`, params.ResponseIndex)
-				data, _ = sjson.Set(data, "content_block.id", fmt.Sprintf("%s-%d", fcName, time.Now().UnixNano()))
+				data, _ = sjson.Set(data, "content_block.id", fmt.Sprintf("%s-%d-%d", fcName, time.Now().UnixNano(), atomic.AddUint64(&toolUseIDCounter, 1)))
 				data, _ = sjson.Set(data, "content_block.name", fcName)
 				output = output + fmt.Sprintf("data: %s\n\n\n", data)
 

internal/translator/antigravity/openai/chat-completions/antigravity_openai_request.go

@@ -48,13 +48,13 @@ func ConvertOpenAIRequestToAntigravity(modelName string, inputRawJSON []byte, _
 			out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.thinkingBudget", -1)
 			out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.include_thoughts", true)
 		case "low":
-			out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.thinkingBudget", util.NormalizeThinkingBudget(modelName, 1024))
+			out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.thinkingBudget", 1024)
 			out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.include_thoughts", true)
 		case "medium":
-			out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.thinkingBudget", util.NormalizeThinkingBudget(modelName, 8192))
+			out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.thinkingBudget", 8192)
 			out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.include_thoughts", true)
 		case "high":
-			out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.thinkingBudget", util.NormalizeThinkingBudget(modelName, 32768))
+			out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.thinkingBudget", 32768)
 			out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.include_thoughts", true)
 		default:
 			out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.thinkingBudget", -1)
@@ -66,15 +66,15 @@ func ConvertOpenAIRequestToAntigravity(modelName string, inputRawJSON []byte, _
 	if !hasOfficialThinking && util.ModelSupportsThinking(modelName) {
 		if tc := gjson.GetBytes(rawJSON, "extra_body.google.thinking_config"); tc.Exists() && tc.IsObject() {
 			var setBudget bool
-			var normalized int
+			var budget int
 
 			if v := tc.Get("thinkingBudget"); v.Exists() {
-				normalized = util.NormalizeThinkingBudget(modelName, int(v.Int()))
-				out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.thinkingBudget", normalized)
+				budget = int(v.Int())
+				out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.thinkingBudget", budget)
 				setBudget = true
 			} else if v := tc.Get("thinking_budget"); v.Exists() {
-				normalized = util.NormalizeThinkingBudget(modelName, int(v.Int()))
-				out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.thinkingBudget", normalized)
+				budget = int(v.Int())
+				out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.thinkingBudget", budget)
 				setBudget = true
 			}
 
@@ -82,7 +82,7 @@ func ConvertOpenAIRequestToAntigravity(modelName string, inputRawJSON []byte, _
 				out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.include_thoughts", v.Bool())
 			} else if v := tc.Get("include_thoughts"); v.Exists() {
 				out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.include_thoughts", v.Bool())
-			} else if setBudget && normalized != 0 {
+			} else if setBudget && budget != 0 {
 				out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.include_thoughts", true)
 			}
 		}
@@ -94,7 +94,7 @@ func ConvertOpenAIRequestToAntigravity(modelName string, inputRawJSON []byte, _
 		if t := gjson.GetBytes(rawJSON, "thinking"); t.Exists() && t.IsObject() {
 			if t.Get("type").String() == "enabled" {
 				if b := t.Get("budget_tokens"); b.Exists() && b.Type == gjson.Number {
-					budget := util.NormalizeThinkingBudget(modelName, int(b.Int()))
+					budget := int(b.Int())
 					out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.thinkingBudget", budget)
 					out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.include_thoughts", true)
 				}
@@ -102,15 +102,6 @@ func ConvertOpenAIRequestToAntigravity(modelName string, inputRawJSON []byte, _
 		}
 	}
 
-	// For gemini-3-pro-preview, always send default thinkingConfig when none specified.
-	// This matches the official Gemini CLI behavior which always sends:
-	// { thinkingBudget: -1, includeThoughts: true }
-	// See: ai-gemini-cli/packages/core/src/config/defaultModelConfigs.ts
-	if !gjson.GetBytes(out, "request.generationConfig.thinkingConfig").Exists() && modelName == "gemini-3-pro-preview" {
-		out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.thinkingBudget", -1)
-		out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.include_thoughts", true)
-	}
-
 	// Temperature/top_p/top_k/max_tokens
 	if tr := gjson.GetBytes(rawJSON, "temperature"); tr.Exists() && tr.Type == gjson.Number {
 		out, _ = sjson.SetBytes(out, "request.generationConfig.temperature", tr.Num)

internal/translator/antigravity/openai/chat-completions/antigravity_openai_response.go

@@ -11,6 +11,7 @@ import (
 	"encoding/json"
 	"fmt"
 	"strings"
+	"sync/atomic"
 	"time"
 
 	. "github.com/router-for-me/CLIProxyAPI/v6/internal/translator/gemini/openai/chat-completions"
@@ -24,6 +25,9 @@ type convertCliResponseToOpenAIChatParams struct {
 	FunctionIndex int
 }
 
+// functionCallIDCounter provides a process-wide unique counter for function call identifiers.
+var functionCallIDCounter uint64
+
 // ConvertAntigravityResponseToOpenAI translates a single chunk of a streaming response from the
 // Gemini CLI API format to the OpenAI Chat Completions streaming format.
 // It processes various Gemini CLI event types and transforms them into OpenAI-compatible JSON responses.
@@ -146,7 +150,7 @@ func ConvertAntigravityResponseToOpenAI(_ context.Context, _ string, originalReq
 
 				functionCallTemplate := `{"id": "","index": 0,"type": "function","function": {"name": "","arguments": ""}}`
 				fcName := functionCallResult.Get("name").String()
-				functionCallTemplate, _ = sjson.Set(functionCallTemplate, "id", fmt.Sprintf("%s-%d", fcName, time.Now().UnixNano()))
+				functionCallTemplate, _ = sjson.Set(functionCallTemplate, "id", fmt.Sprintf("%s-%d-%d", fcName, time.Now().UnixNano(), atomic.AddUint64(&functionCallIDCounter, 1)))
 				functionCallTemplate, _ = sjson.Set(functionCallTemplate, "index", functionCallIndex)
 				functionCallTemplate, _ = sjson.Set(functionCallTemplate, "function.name", fcName)
 				if fcArgsResult := functionCallResult.Get("args"); fcArgsResult.Exists() {

internal/translator/gemini-cli/claude/gemini-cli_claude_request.go

@@ -165,7 +165,6 @@ func ConvertClaudeRequestToCLI(modelName string, inputRawJSON []byte, _ bool) []
 		if t.Get("type").String() == "enabled" {
 			if b := t.Get("budget_tokens"); b.Exists() && b.Type == gjson.Number {
 				budget := int(b.Int())
-				budget = util.NormalizeThinkingBudget(modelName, budget)
 				out, _ = sjson.Set(out, "request.generationConfig.thinkingConfig.thinkingBudget", budget)
 				out, _ = sjson.Set(out, "request.generationConfig.thinkingConfig.include_thoughts", true)
 			}

internal/translator/gemini-cli/claude/gemini-cli_claude_response.go

@@ -12,6 +12,7 @@ import (
 	"encoding/json"
 	"fmt"
 	"strings"
+	"sync/atomic"
 	"time"
 
 	"github.com/tidwall/gjson"
@@ -27,6 +28,9 @@ type Params struct {
 	ResponseIndex    int  // Index counter for content blocks in the streaming response
 }
 
+// toolUseIDCounter provides a process-wide unique counter for tool use identifiers.
+var toolUseIDCounter uint64
+
 // ConvertGeminiCLIResponseToClaude performs sophisticated streaming response format conversion.
 // This function implements a complex state machine that translates backend client responses
 // into Claude Code-compatible Server-Sent Events (SSE) format. It manages different response types
@@ -197,7 +201,7 @@ func ConvertGeminiCLIResponseToClaude(_ context.Context, _ string, originalReque
 
 				// Create the tool use block with unique ID and function details
 				data := fmt.Sprintf(`{"type":"content_block_start","index":%d,"content_block":{"type":"tool_use","id":"","name":"","input":{}}}`, (*param).(*Params).ResponseIndex)
-				data, _ = sjson.Set(data, "content_block.id", fmt.Sprintf("%s-%d", fcName, time.Now().UnixNano()))
+				data, _ = sjson.Set(data, "content_block.id", fmt.Sprintf("%s-%d-%d", fcName, time.Now().UnixNano(), atomic.AddUint64(&toolUseIDCounter, 1)))
 				data, _ = sjson.Set(data, "content_block.name", fcName)
 				output = output + fmt.Sprintf("data: %s\n\n\n", data)
 

internal/translator/gemini-cli/openai/chat-completions/gemini-cli_openai_request.go

@@ -48,13 +48,13 @@ func ConvertOpenAIRequestToGeminiCLI(modelName string, inputRawJSON []byte, _ bo
 			out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.thinkingBudget", -1)
 			out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.include_thoughts", true)
 		case "low":
-			out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.thinkingBudget", util.NormalizeThinkingBudget(modelName, 1024))
+			out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.thinkingBudget", 1024)
 			out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.include_thoughts", true)
 		case "medium":
-			out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.thinkingBudget", util.NormalizeThinkingBudget(modelName, 8192))
+			out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.thinkingBudget", 8192)
 			out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.include_thoughts", true)
 		case "high":
-			out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.thinkingBudget", util.NormalizeThinkingBudget(modelName, 32768))
+			out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.thinkingBudget", 32768)
 			out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.include_thoughts", true)
 		default:
 			out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.thinkingBudget", -1)
@@ -66,15 +66,15 @@ func ConvertOpenAIRequestToGeminiCLI(modelName string, inputRawJSON []byte, _ bo
 	if !hasOfficialThinking && util.ModelSupportsThinking(modelName) {
 		if tc := gjson.GetBytes(rawJSON, "extra_body.google.thinking_config"); tc.Exists() && tc.IsObject() {
 			var setBudget bool
-			var normalized int
+			var budget int
 
 			if v := tc.Get("thinkingBudget"); v.Exists() {
-				normalized = util.NormalizeThinkingBudget(modelName, int(v.Int()))
-				out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.thinkingBudget", normalized)
+				budget = int(v.Int())
+				out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.thinkingBudget", budget)
 				setBudget = true
 			} else if v := tc.Get("thinking_budget"); v.Exists() {
-				normalized = util.NormalizeThinkingBudget(modelName, int(v.Int()))
-				out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.thinkingBudget", normalized)
+				budget = int(v.Int())
+				out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.thinkingBudget", budget)
 				setBudget = true
 			}
 
@@ -82,21 +82,12 @@ func ConvertOpenAIRequestToGeminiCLI(modelName string, inputRawJSON []byte, _ bo
 				out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.include_thoughts", v.Bool())
 			} else if v := tc.Get("include_thoughts"); v.Exists() {
 				out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.include_thoughts", v.Bool())
-			} else if setBudget && normalized != 0 {
+			} else if setBudget && budget != 0 {
 				out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.include_thoughts", true)
 			}
 		}
 	}
 
-	// For gemini-3-pro-preview, always send default thinkingConfig when none specified.
-	// This matches the official Gemini CLI behavior which always sends:
-	// { thinkingBudget: -1, includeThoughts: true }
-	// See: ai-gemini-cli/packages/core/src/config/defaultModelConfigs.ts
-	if !gjson.GetBytes(out, "request.generationConfig.thinkingConfig").Exists() && modelName == "gemini-3-pro-preview" {
-		out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.thinkingBudget", -1)
-		out, _ = sjson.SetBytes(out, "request.generationConfig.thinkingConfig.include_thoughts", true)
-	}
-
 	// Temperature/top_p/top_k
 	if tr := gjson.GetBytes(rawJSON, "temperature"); tr.Exists() && tr.Type == gjson.Number {
 		out, _ = sjson.SetBytes(out, "request.generationConfig.temperature", tr.Num)

internal/translator/gemini-cli/openai/chat-completions/gemini-cli_openai_response.go

@@ -11,6 +11,7 @@ import (
 	"encoding/json"
 	"fmt"
 	"strings"
+	"sync/atomic"
 	"time"
 
 	. "github.com/router-for-me/CLIProxyAPI/v6/internal/translator/gemini/openai/chat-completions"
@@ -24,6 +25,9 @@ type convertCliResponseToOpenAIChatParams struct {
 	FunctionIndex int
 }
 
+// functionCallIDCounter provides a process-wide unique counter for function call identifiers.
+var functionCallIDCounter uint64
+
 // ConvertCliResponseToOpenAI translates a single chunk of a streaming response from the
 // Gemini CLI API format to the OpenAI Chat Completions streaming format.
 // It processes various Gemini CLI event types and transforms them into OpenAI-compatible JSON responses.
@@ -146,7 +150,7 @@ func ConvertCliResponseToOpenAI(_ context.Context, _ string, originalRequestRawJ
 
 				functionCallTemplate := `{"id": "","index": 0,"type": "function","function": {"name": "","arguments": ""}}`
 				fcName := functionCallResult.Get("name").String()
-				functionCallTemplate, _ = sjson.Set(functionCallTemplate, "id", fmt.Sprintf("%s-%d", fcName, time.Now().UnixNano()))
+				functionCallTemplate, _ = sjson.Set(functionCallTemplate, "id", fmt.Sprintf("%s-%d-%d", fcName, time.Now().UnixNano(), atomic.AddUint64(&functionCallIDCounter, 1)))
 				functionCallTemplate, _ = sjson.Set(functionCallTemplate, "index", functionCallIndex)
 				functionCallTemplate, _ = sjson.Set(functionCallTemplate, "function.name", fcName)
 				if fcArgsResult := functionCallResult.Get("args"); fcArgsResult.Exists() {

internal/translator/gemini/claude/gemini_claude_request.go

@@ -158,7 +158,6 @@ func ConvertClaudeRequestToGemini(modelName string, inputRawJSON []byte, _ bool)
 		if t.Get("type").String() == "enabled" {
 			if b := t.Get("budget_tokens"); b.Exists() && b.Type == gjson.Number {
 				budget := int(b.Int())
-				budget = util.NormalizeThinkingBudget(modelName, budget)
 				out, _ = sjson.Set(out, "generationConfig.thinkingConfig.thinkingBudget", budget)
 				out, _ = sjson.Set(out, "generationConfig.thinkingConfig.include_thoughts", true)
 			}

internal/translator/gemini/claude/gemini_claude_response.go

@@ -12,6 +12,7 @@ import (
 	"encoding/json"
 	"fmt"
 	"strings"
+	"sync/atomic"
 	"time"
 
 	"github.com/tidwall/gjson"
@@ -26,6 +27,9 @@ type Params struct {
 	ResponseIndex    int
 }
 
+// toolUseIDCounter provides a process-wide unique counter for tool use identifiers.
+var toolUseIDCounter uint64
+
 // ConvertGeminiResponseToClaude performs sophisticated streaming response format conversion.
 // This function implements a complex state machine that translates backend client responses
 // into Claude-compatible Server-Sent Events (SSE) format. It manages different response types
@@ -197,7 +201,7 @@ func ConvertGeminiResponseToClaude(_ context.Context, _ string, originalRequestR
 
 				// Create the tool use block with unique ID and function details
 				data := fmt.Sprintf(`{"type":"content_block_start","index":%d,"content_block":{"type":"tool_use","id":"","name":"","input":{}}}`, (*param).(*Params).ResponseIndex)
-				data, _ = sjson.Set(data, "content_block.id", fmt.Sprintf("%s-%d", fcName, time.Now().UnixNano()))
+				data, _ = sjson.Set(data, "content_block.id", fmt.Sprintf("%s-%d-%d", fcName, time.Now().UnixNano(), atomic.AddUint64(&toolUseIDCounter, 1)))
 				data, _ = sjson.Set(data, "content_block.name", fcName)
 				output = output + fmt.Sprintf("data: %s\n\n\n", data)
 

internal/translator/gemini/openai/chat-completions/gemini_openai_request.go

@@ -48,13 +48,13 @@ func ConvertOpenAIRequestToGemini(modelName string, inputRawJSON []byte, _ bool)
 			out, _ = sjson.SetBytes(out, "generationConfig.thinkingConfig.thinkingBudget", -1)
 			out, _ = sjson.SetBytes(out, "generationConfig.thinkingConfig.include_thoughts", true)
 		case "low":
-			out, _ = sjson.SetBytes(out, "generationConfig.thinkingConfig.thinkingBudget", util.NormalizeThinkingBudget(modelName, 1024))
+			out, _ = sjson.SetBytes(out, "generationConfig.thinkingConfig.thinkingBudget", 1024)
 			out, _ = sjson.SetBytes(out, "generationConfig.thinkingConfig.include_thoughts", true)
 		case "medium":
-			out, _ = sjson.SetBytes(out, "generationConfig.thinkingConfig.thinkingBudget", util.NormalizeThinkingBudget(modelName, 8192))
+			out, _ = sjson.SetBytes(out, "generationConfig.thinkingConfig.thinkingBudget", 8192)
 			out, _ = sjson.SetBytes(out, "generationConfig.thinkingConfig.include_thoughts", true)
 		case "high":
-			out, _ = sjson.SetBytes(out, "generationConfig.thinkingConfig.thinkingBudget", util.NormalizeThinkingBudget(modelName, 32768))
+			out, _ = sjson.SetBytes(out, "generationConfig.thinkingConfig.thinkingBudget", 32768)
 			out, _ = sjson.SetBytes(out, "generationConfig.thinkingConfig.include_thoughts", true)
 		default:
 			out, _ = sjson.SetBytes(out, "generationConfig.thinkingConfig.thinkingBudget", -1)
@@ -66,15 +66,15 @@ func ConvertOpenAIRequestToGemini(modelName string, inputRawJSON []byte, _ bool)
 	if !hasOfficialThinking && util.ModelSupportsThinking(modelName) {
 		if tc := gjson.GetBytes(rawJSON, "extra_body.google.thinking_config"); tc.Exists() && tc.IsObject() {
 			var setBudget bool
-			var normalized int
+			var budget int
 
 			if v := tc.Get("thinkingBudget"); v.Exists() {
-				normalized = util.NormalizeThinkingBudget(modelName, int(v.Int()))
-				out, _ = sjson.SetBytes(out, "generationConfig.thinkingConfig.thinkingBudget", normalized)
+				budget = int(v.Int())
+				out, _ = sjson.SetBytes(out, "generationConfig.thinkingConfig.thinkingBudget", budget)
 				setBudget = true
 			} else if v := tc.Get("thinking_budget"); v.Exists() {
-				normalized = util.NormalizeThinkingBudget(modelName, int(v.Int()))
-				out, _ = sjson.SetBytes(out, "generationConfig.thinkingConfig.thinkingBudget", normalized)
+				budget = int(v.Int())
+				out, _ = sjson.SetBytes(out, "generationConfig.thinkingConfig.thinkingBudget", budget)
 				setBudget = true
 			}
 
@@ -82,7 +82,7 @@ func ConvertOpenAIRequestToGemini(modelName string, inputRawJSON []byte, _ bool)
 				out, _ = sjson.SetBytes(out, "generationConfig.thinkingConfig.include_thoughts", v.Bool())
 			} else if v := tc.Get("include_thoughts"); v.Exists() {
 				out, _ = sjson.SetBytes(out, "generationConfig.thinkingConfig.include_thoughts", v.Bool())
-			} else if setBudget && normalized != 0 {
+			} else if setBudget && budget != 0 {
 				out, _ = sjson.SetBytes(out, "generationConfig.thinkingConfig.include_thoughts", true)
 			}
 		}

internal/translator/gemini/openai/chat-completions/gemini_openai_response.go

@@ -11,6 +11,7 @@ import (
 	"encoding/json"
 	"fmt"
 	"strings"
+	"sync/atomic"
 	"time"
 
 	"github.com/tidwall/gjson"
@@ -23,6 +24,9 @@ type convertGeminiResponseToOpenAIChatParams struct {
 	FunctionIndex int
 }
 
+// functionCallIDCounter provides a process-wide unique counter for function call identifiers.
+var functionCallIDCounter uint64
+
 // ConvertGeminiResponseToOpenAI translates a single chunk of a streaming response from the
 // Gemini API format to the OpenAI Chat Completions streaming format.
 // It processes various Gemini event types and transforms them into OpenAI-compatible JSON responses.
@@ -148,7 +152,7 @@ func ConvertGeminiResponseToOpenAI(_ context.Context, _ string, originalRequestR
 
 				functionCallTemplate := `{"id": "","index": 0,"type": "function","function": {"name": "","arguments": ""}}`
 				fcName := functionCallResult.Get("name").String()
-				functionCallTemplate, _ = sjson.Set(functionCallTemplate, "id", fmt.Sprintf("%s-%d", fcName, time.Now().UnixNano()))
+				functionCallTemplate, _ = sjson.Set(functionCallTemplate, "id", fmt.Sprintf("%s-%d-%d", fcName, time.Now().UnixNano(), atomic.AddUint64(&functionCallIDCounter, 1)))
 				functionCallTemplate, _ = sjson.Set(functionCallTemplate, "index", functionCallIndex)
 				functionCallTemplate, _ = sjson.Set(functionCallTemplate, "function.name", fcName)
 				if fcArgsResult := functionCallResult.Get("args"); fcArgsResult.Exists() {
@@ -281,7 +285,7 @@ func ConvertGeminiResponseToOpenAINonStream(_ context.Context, _ string, origina
 				}
 				functionCallItemTemplate := `{"id": "","type": "function","function": {"name": "","arguments": ""}}`
 				fcName := functionCallResult.Get("name").String()
-				functionCallItemTemplate, _ = sjson.Set(functionCallItemTemplate, "id", fmt.Sprintf("%s-%d", fcName, time.Now().UnixNano()))
+				functionCallItemTemplate, _ = sjson.Set(functionCallItemTemplate, "id", fmt.Sprintf("%s-%d-%d", fcName, time.Now().UnixNano(), atomic.AddUint64(&functionCallIDCounter, 1)))
 				functionCallItemTemplate, _ = sjson.Set(functionCallItemTemplate, "function.name", fcName)
 				if fcArgsResult := functionCallResult.Get("args"); fcArgsResult.Exists() {
 					functionCallItemTemplate, _ = sjson.Set(functionCallItemTemplate, "function.arguments", fcArgsResult.Raw)

internal/translator/gemini/openai/responses/gemini_openai-responses_request.go

@@ -400,16 +400,16 @@ func ConvertOpenAIResponsesRequestToGemini(modelName string, inputRawJSON []byte
 			out, _ = sjson.Set(out, "generationConfig.thinkingConfig.thinkingBudget", -1)
 			out, _ = sjson.Set(out, "generationConfig.thinkingConfig.include_thoughts", true)
 		case "minimal":
-			out, _ = sjson.Set(out, "generationConfig.thinkingConfig.thinkingBudget", util.NormalizeThinkingBudget(modelName, 1024))
+			out, _ = sjson.Set(out, "generationConfig.thinkingConfig.thinkingBudget", 1024)
 			out, _ = sjson.Set(out, "generationConfig.thinkingConfig.include_thoughts", true)
 		case "low":
-			out, _ = sjson.Set(out, "generationConfig.thinkingConfig.thinkingBudget", util.NormalizeThinkingBudget(modelName, 4096))
+			out, _ = sjson.Set(out, "generationConfig.thinkingConfig.thinkingBudget", 4096)
 			out, _ = sjson.Set(out, "generationConfig.thinkingConfig.include_thoughts", true)
 		case "medium":
-			out, _ = sjson.Set(out, "generationConfig.thinkingConfig.thinkingBudget", util.NormalizeThinkingBudget(modelName, 8192))
+			out, _ = sjson.Set(out, "generationConfig.thinkingConfig.thinkingBudget", 8192)
 			out, _ = sjson.Set(out, "generationConfig.thinkingConfig.include_thoughts", true)
 		case "high":
-			out, _ = sjson.Set(out, "generationConfig.thinkingConfig.thinkingBudget", util.NormalizeThinkingBudget(modelName, 32768))
+			out, _ = sjson.Set(out, "generationConfig.thinkingConfig.thinkingBudget", 32768)
 			out, _ = sjson.Set(out, "generationConfig.thinkingConfig.include_thoughts", true)
 		default:
 			out, _ = sjson.Set(out, "generationConfig.thinkingConfig.thinkingBudget", -1)
@@ -421,32 +421,22 @@ func ConvertOpenAIResponsesRequestToGemini(modelName string, inputRawJSON []byte
 	if !hasOfficialThinking && util.ModelSupportsThinking(modelName) {
 		if tc := root.Get("extra_body.google.thinking_config"); tc.Exists() && tc.IsObject() {
 			var setBudget bool
-			var normalized int
+			var budget int
 			if v := tc.Get("thinking_budget"); v.Exists() {
-				normalized = util.NormalizeThinkingBudget(modelName, int(v.Int()))
-				out, _ = sjson.Set(out, "generationConfig.thinkingConfig.thinkingBudget", normalized)
+				budget = int(v.Int())
+				out, _ = sjson.Set(out, "generationConfig.thinkingConfig.thinkingBudget", budget)
 				setBudget = true
 			}
 			if v := tc.Get("include_thoughts"); v.Exists() {
 				out, _ = sjson.Set(out, "generationConfig.thinkingConfig.include_thoughts", v.Bool())
 			} else if setBudget {
-				if normalized != 0 {
+				if budget != 0 {
 					out, _ = sjson.Set(out, "generationConfig.thinkingConfig.include_thoughts", true)
 				}
 			}
 		}
 	}
 
-	// For gemini-3-pro-preview, always send default thinkingConfig when none specified.
-	// This matches the official Gemini CLI behavior which always sends:
-	// { thinkingBudget: -1, includeThoughts: true }
-	// See: ai-gemini-cli/packages/core/src/config/defaultModelConfigs.ts
-	if !gjson.Get(out, "generationConfig.thinkingConfig").Exists() && modelName == "gemini-3-pro-preview" {
-		out, _ = sjson.Set(out, "generationConfig.thinkingConfig.thinkingBudget", -1)
-		out, _ = sjson.Set(out, "generationConfig.thinkingConfig.include_thoughts", true)
-		// log.Debugf("Applied default thinkingConfig for gemini-3-pro-preview (matches Gemini CLI): thinkingBudget=-1, include_thoughts=true")
-	}
-
 	result := []byte(out)
 	result = common.AttachDefaultSafetySettings(result, "safetySettings")
 	return result

internal/translator/gemini/openai/responses/gemini_openai-responses_response.go

@@ -5,6 +5,7 @@ import (
 	"context"
 	"fmt"
 	"strings"
+	"sync/atomic"
 	"time"
 
 	"github.com/tidwall/gjson"
@@ -37,6 +38,12 @@ type geminiToResponsesState struct {
 	FuncCallIDs map[int]string
 }
 
+// responseIDCounter provides a process-wide unique counter for synthesized response identifiers.
+var responseIDCounter uint64
+
+// funcCallIDCounter provides a process-wide unique counter for function call identifiers.
+var funcCallIDCounter uint64
+
 func emitEvent(event string, payload string) string {
 	return fmt.Sprintf("event: %s\ndata: %s", event, payload)
 }
@@ -205,7 +212,7 @@ func ConvertGeminiResponseToOpenAIResponses(_ context.Context, modelName string,
 					st.FuncArgsBuf[idx] = &strings.Builder{}
 				}
 				if st.FuncCallIDs[idx] == "" {
-					st.FuncCallIDs[idx] = fmt.Sprintf("call_%d", time.Now().UnixNano())
+					st.FuncCallIDs[idx] = fmt.Sprintf("call_%d_%d", time.Now().UnixNano(), atomic.AddUint64(&funcCallIDCounter, 1))
 				}
 				st.FuncNames[idx] = name
 
@@ -464,7 +471,7 @@ func ConvertGeminiResponseToOpenAIResponsesNonStream(_ context.Context, _ string
 	// id: prefer provider responseId, otherwise synthesize
 	id := root.Get("responseId").String()
 	if id == "" {
-		id = fmt.Sprintf("resp_%x", time.Now().UnixNano())
+		id = fmt.Sprintf("resp_%x_%d", time.Now().UnixNano(), atomic.AddUint64(&responseIDCounter, 1))
 	}
 	// Normalize to response-style id (prefix resp_ if missing)
 	if !strings.HasPrefix(id, "resp_") {
@@ -575,7 +582,7 @@ func ConvertGeminiResponseToOpenAIResponsesNonStream(_ context.Context, _ string
 			if fc := p.Get("functionCall"); fc.Exists() {
 				name := fc.Get("name").String()
 				args := fc.Get("args")
-				callID := fmt.Sprintf("call_%x", time.Now().UnixNano())
+				callID := fmt.Sprintf("call_%x_%d", time.Now().UnixNano(), atomic.AddUint64(&funcCallIDCounter, 1))
 				outputs = append(outputs, map[string]interface{}{
 					"id":     fmt.Sprintf("fc_%s", callID),
 					"type":   "function_call",

internal/translator/openai/openai/responses/openai_openai-responses_response.go

@@ -5,6 +5,7 @@ import (
 	"context"
 	"fmt"
 	"strings"
+	"sync/atomic"
 	"time"
 
 	"github.com/tidwall/gjson"
@@ -41,6 +42,9 @@ type oaiToResponsesState struct {
 	UsageSeen        bool
 }
 
+// responseIDCounter provides a process-wide unique counter for synthesized response identifiers.
+var responseIDCounter uint64
+
 func emitRespEvent(event string, payload string) string {
 	return fmt.Sprintf("event: %s\ndata: %s", event, payload)
 }
@@ -590,7 +594,7 @@ func ConvertOpenAIChatCompletionsResponseToOpenAIResponsesNonStream(_ context.Co
 	// id: use provider id if present, otherwise synthesize
 	id := root.Get("id").String()
 	if id == "" {
-		id = fmt.Sprintf("resp_%x", time.Now().UnixNano())
+		id = fmt.Sprintf("resp_%x_%d", time.Now().UnixNano(), atomic.AddUint64(&responseIDCounter, 1))
 	}
 	resp, _ = sjson.Set(resp, "id", id)
 

internal/util/gemini_thinking.go

@@ -207,6 +207,47 @@ func GeminiThinkingFromMetadata(metadata map[string]any) (*int, *bool, bool) {
 	return budgetPtr, includePtr, matched
 }
 
+// modelsWithDefaultThinking lists models that should have thinking enabled by default
+// when no explicit thinkingConfig is provided.
+var modelsWithDefaultThinking = map[string]bool{
+	"gemini-3-pro-preview": true,
+}
+
+// ModelHasDefaultThinking returns true if the model should have thinking enabled by default.
+func ModelHasDefaultThinking(model string) bool {
+	return modelsWithDefaultThinking[model]
+}
+
+// ApplyDefaultThinkingIfNeeded injects default thinkingConfig for models that require it.
+// For standard Gemini API format (generationConfig.thinkingConfig path).
+// Returns the modified body if thinkingConfig was added, otherwise returns the original.
+func ApplyDefaultThinkingIfNeeded(model string, body []byte) []byte {
+	if !ModelHasDefaultThinking(model) {
+		return body
+	}
+	if gjson.GetBytes(body, "generationConfig.thinkingConfig").Exists() {
+		return body
+	}
+	updated, _ := sjson.SetBytes(body, "generationConfig.thinkingConfig.thinkingBudget", -1)
+	updated, _ = sjson.SetBytes(updated, "generationConfig.thinkingConfig.include_thoughts", true)
+	return updated
+}
+
+// ApplyDefaultThinkingIfNeededCLI injects default thinkingConfig for models that require it.
+// For Gemini CLI API format (request.generationConfig.thinkingConfig path).
+// Returns the modified body if thinkingConfig was added, otherwise returns the original.
+func ApplyDefaultThinkingIfNeededCLI(model string, body []byte) []byte {
+	if !ModelHasDefaultThinking(model) {
+		return body
+	}
+	if gjson.GetBytes(body, "request.generationConfig.thinkingConfig").Exists() {
+		return body
+	}
+	updated, _ := sjson.SetBytes(body, "request.generationConfig.thinkingConfig.thinkingBudget", -1)
+	updated, _ = sjson.SetBytes(updated, "request.generationConfig.thinkingConfig.include_thoughts", true)
+	return updated
+}
+
 // StripThinkingConfigIfUnsupported removes thinkingConfig from the request body
 // when the target model does not advertise Thinking capability. It cleans both
 // standard Gemini and Gemini CLI JSON envelopes. This acts as a final safety net
@@ -223,6 +264,32 @@ func StripThinkingConfigIfUnsupported(model string, body []byte) []byte {
 	return updated
 }
 
+// NormalizeGeminiThinkingBudget normalizes the thinkingBudget value in a standard Gemini
+// request body (generationConfig.thinkingConfig.thinkingBudget path).
+func NormalizeGeminiThinkingBudget(model string, body []byte) []byte {
+	const budgetPath = "generationConfig.thinkingConfig.thinkingBudget"
+	budget := gjson.GetBytes(body, budgetPath)
+	if !budget.Exists() {
+		return body
+	}
+	normalized := NormalizeThinkingBudget(model, int(budget.Int()))
+	updated, _ := sjson.SetBytes(body, budgetPath, normalized)
+	return updated
+}
+
+// NormalizeGeminiCLIThinkingBudget normalizes the thinkingBudget value in a Gemini CLI
+// request body (request.generationConfig.thinkingConfig.thinkingBudget path).
+func NormalizeGeminiCLIThinkingBudget(model string, body []byte) []byte {
+	const budgetPath = "request.generationConfig.thinkingConfig.thinkingBudget"
+	budget := gjson.GetBytes(body, budgetPath)
+	if !budget.Exists() {
+		return body
+	}
+	normalized := NormalizeThinkingBudget(model, int(budget.Int()))
+	updated, _ := sjson.SetBytes(body, budgetPath, normalized)
+	return updated
+}
+
 // ConvertThinkingLevelToBudget checks for "generationConfig.thinkingConfig.thinkingLevel"
 // and converts it to "thinkingBudget".
 // "high" -> 32768

sdk/api/handlers/gemini/gemini_handlers.go

@@ -48,8 +48,24 @@ func (h *GeminiAPIHandler) Models() []map[string]any {
 // GeminiModels handles the Gemini models listing endpoint.
 // It returns a JSON response containing available Gemini models and their specifications.
 func (h *GeminiAPIHandler) GeminiModels(c *gin.Context) {
+	rawModels := h.Models()
+	normalizedModels := make([]map[string]any, 0, len(rawModels))
+	defaultMethods := []string{"generateContent"}
+	for _, model := range rawModels {
+		normalizedModel := make(map[string]any, len(model))
+		for k, v := range model {
+			normalizedModel[k] = v
+		}
+		if name, ok := normalizedModel["name"].(string); ok && name != "" && !strings.HasPrefix(name, "models/") {
+			normalizedModel["name"] = "models/" + name
+		}
+		if _, ok := normalizedModel["supportedGenerationMethods"]; !ok {
+			normalizedModel["supportedGenerationMethods"] = defaultMethods
+		}
+		normalizedModels = append(normalizedModels, normalizedModel)
+	}
 	c.JSON(http.StatusOK, gin.H{
-		"models": h.Models(),
+		"models": normalizedModels,
 	})
 }
 

sdk/cliproxy/service.go

@@ -779,7 +779,7 @@ func (s *Service) registerModelsForAuth(a *coreauth.Auth) {
 							Created:     time.Now().Unix(),
 							OwnedBy:     compat.Name,
 							Type:        "openai-compatibility",
-							DisplayName: m.Name,
+							DisplayName: modelID,
 						})
 					}
 					// Register and return