Fixed: #337

**fix(executor): replace redundant commented code with `checkSystemInstructions` helper**

- Replaced commented-out `sjson.SetRawBytes` lines with the new `checkSystemInstructions` function.
- Centralized system instruction handling for better code clarity and reuse.
- Ensured consistent logic for managing `system` field across Claude executor flows.

internal/api/middleware/request_logging.go

@@ -6,6 +6,7 @@ package middleware
 import (
 	"bytes"
 	"io"
+	"net/http"
 	"strings"
 
 	"github.com/gin-gonic/gin"
@@ -24,6 +25,11 @@ func RequestLoggingMiddleware(logger logging.RequestLogger) gin.HandlerFunc {
 			return
 		}
 
+		if c.Request.Method == http.MethodGet {
+			c.Next()
+			return
+		}
+
 		path := c.Request.URL.Path
 		if !shouldLogRequest(path) {
 			c.Next()

internal/registry/model_definitions.go

@@ -23,6 +23,60 @@ func GetClaudeModels() []*ModelInfo {
 			Type:        "claude",
 			DisplayName: "Claude 4.5 Sonnet",
 		},
+		{
+			ID:          "claude-sonnet-4-5-thinking",
+			Object:      "model",
+			Created:     1759104000, // 2025-09-29
+			OwnedBy:     "anthropic",
+			Type:        "claude",
+			DisplayName: "Claude 4.5 Sonnet Thinking",
+			Thinking:    &ThinkingSupport{Min: 1024, Max: 100000, ZeroAllowed: false, DynamicAllowed: true},
+		},
+		{
+			ID:          "gemini-claude-sonnet-4-5-thinking",
+			Object:      "model",
+			Created:     1759104000, // 2025-09-29
+			OwnedBy:     "anthropic",
+			Type:        "claude",
+			DisplayName: "Gemini Claude 4.5 Sonnet Thinking",
+			Thinking:    &ThinkingSupport{Min: 1024, Max: 100000, ZeroAllowed: false, DynamicAllowed: true},
+		},
+		{
+			ID:          "claude-opus-4-5-thinking",
+			Object:      "model",
+			Created:     1761955200, // 2025-11-01
+			OwnedBy:     "anthropic",
+			Type:        "claude",
+			DisplayName: "Claude 4.5 Opus Thinking",
+			Thinking:    &ThinkingSupport{Min: 1024, Max: 100000, ZeroAllowed: false, DynamicAllowed: true},
+		},
+		{
+			ID:          "claude-opus-4-5-thinking-low",
+			Object:      "model",
+			Created:     1761955200, // 2025-11-01
+			OwnedBy:     "anthropic",
+			Type:        "claude",
+			DisplayName: "Claude 4.5 Opus Thinking Low",
+			Thinking:    &ThinkingSupport{Min: 1024, Max: 100000, ZeroAllowed: false, DynamicAllowed: true},
+		},
+		{
+			ID:          "claude-opus-4-5-thinking-medium",
+			Object:      "model",
+			Created:     1761955200, // 2025-11-01
+			OwnedBy:     "anthropic",
+			Type:        "claude",
+			DisplayName: "Claude 4.5 Opus Thinking Medium",
+			Thinking:    &ThinkingSupport{Min: 1024, Max: 100000, ZeroAllowed: false, DynamicAllowed: true},
+		},
+		{
+			ID:          "claude-opus-4-5-thinking-high",
+			Object:      "model",
+			Created:     1761955200, // 2025-11-01
+			OwnedBy:     "anthropic",
+			Type:        "claude",
+			DisplayName: "Claude 4.5 Opus Thinking High",
+			Thinking:    &ThinkingSupport{Min: 1024, Max: 100000, ZeroAllowed: false, DynamicAllowed: true},
+		},
 		{
 			ID:                  "claude-opus-4-5-20251101",
 			Object:              "model",

internal/runtime/executor/antigravity_executor.go

@@ -365,13 +365,27 @@ func FetchAntigravityModels(ctx context.Context, auth *cliproxyauth.Auth, cfg *c
 		for id := range result.Map() {
 			id = modelName2Alias(id)
 			if id != "" {
-				models = append(models, &registry.ModelInfo{
+				modelInfo := &registry.ModelInfo{
-					ID:      id,
+					ID:          id,
-					Object:  "model",
+					Name:        id,
-					Created: now,
+					Description: id,
-					OwnedBy: antigravityAuthType,
+					DisplayName: id,
-					Type:    antigravityAuthType,
+					Version:     id,
-				})
+					Object:      "model",
+					Created:     now,
+					OwnedBy:     antigravityAuthType,
+					Type:        antigravityAuthType,
+				}
+				// Add Thinking support for thinking models
+				if strings.HasSuffix(id, "-thinking") || strings.Contains(id, "-thinking-") {
+					modelInfo.Thinking = &registry.ThinkingSupport{
+						Min:            1024,
+						Max:            100000,
+						ZeroAllowed:    false,
+						DynamicAllowed: true,
+					}
+				}
+				models = append(models, modelInfo)
 			}
 		}
 		return models

internal/runtime/executor/claude_executor.go

@@ -58,18 +58,24 @@ func (e *ClaudeExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, r
 		body, _ = sjson.SetBytes(body, "model", modelOverride)
 		modelForUpstream = modelOverride
 	}
+	// Inject thinking config based on model suffix for thinking variants
+	body = e.injectThinkingConfig(req.Model, body)
 
 	if !strings.HasPrefix(modelForUpstream, "claude-3-5-haiku") {
-		body, _ = sjson.SetRawBytes(body, "system", []byte(misc.ClaudeCodeInstructions))
+		body = checkSystemInstructions(body)
 	}
 	body = applyPayloadConfig(e.cfg, req.Model, body)
 
+	// Extract betas from body and convert to header
+	var extraBetas []string
+	extraBetas, body = extractAndRemoveBetas(body)
+
 	url := fmt.Sprintf("%s/v1/messages?beta=true", baseURL)
 	httpReq, err := http.NewRequestWithContext(ctx, http.MethodPost, url, bytes.NewReader(body))
 	if err != nil {
 		return resp, err
 	}
-	applyClaudeHeaders(httpReq, auth, apiKey, false)
+	applyClaudeHeaders(httpReq, auth, apiKey, false, extraBetas)
 	var authID, authLabel, authType, authValue string
 	if auth != nil {
 		authID = auth.ID
@@ -154,15 +160,21 @@ func (e *ClaudeExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.A
 	if modelOverride := e.resolveUpstreamModel(req.Model, auth); modelOverride != "" {
 		body, _ = sjson.SetBytes(body, "model", modelOverride)
 	}
-	body, _ = sjson.SetRawBytes(body, "system", []byte(misc.ClaudeCodeInstructions))
+	// Inject thinking config based on model suffix for thinking variants
+	body = e.injectThinkingConfig(req.Model, body)
+	body = checkSystemInstructions(body)
 	body = applyPayloadConfig(e.cfg, req.Model, body)
 
+	// Extract betas from body and convert to header
+	var extraBetas []string
+	extraBetas, body = extractAndRemoveBetas(body)
+
 	url := fmt.Sprintf("%s/v1/messages?beta=true", baseURL)
 	httpReq, err := http.NewRequestWithContext(ctx, http.MethodPost, url, bytes.NewReader(body))
 	if err != nil {
 		return nil, err
 	}
-	applyClaudeHeaders(httpReq, auth, apiKey, true)
+	applyClaudeHeaders(httpReq, auth, apiKey, true, extraBetas)
 	var authID, authLabel, authType, authValue string
 	if auth != nil {
 		authID = auth.ID
@@ -283,15 +295,19 @@ func (e *ClaudeExecutor) CountTokens(ctx context.Context, auth *cliproxyauth.Aut
 	}
 
 	if !strings.HasPrefix(modelForUpstream, "claude-3-5-haiku") {
-		body, _ = sjson.SetRawBytes(body, "system", []byte(misc.ClaudeCodeInstructions))
+		body = checkSystemInstructions(body)
 	}
 
+	// Extract betas from body and convert to header (for count_tokens too)
+	var extraBetas []string
+	extraBetas, body = extractAndRemoveBetas(body)
+
 	url := fmt.Sprintf("%s/v1/messages/count_tokens?beta=true", baseURL)
 	httpReq, err := http.NewRequestWithContext(ctx, http.MethodPost, url, bytes.NewReader(body))
 	if err != nil {
 		return cliproxyexecutor.Response{}, err
 	}
-	applyClaudeHeaders(httpReq, auth, apiKey, false)
+	applyClaudeHeaders(httpReq, auth, apiKey, false, extraBetas)
 	var authID, authLabel, authType, authValue string
 	if auth != nil {
 		authID = auth.ID
@@ -383,10 +399,65 @@ func (e *ClaudeExecutor) Refresh(ctx context.Context, auth *cliproxyauth.Auth) (
 	return auth, nil
 }
 
+// extractAndRemoveBetas extracts the "betas" array from the body and removes it.
+// Returns the extracted betas as a string slice and the modified body.
+func extractAndRemoveBetas(body []byte) ([]string, []byte) {
+	betasResult := gjson.GetBytes(body, "betas")
+	if !betasResult.Exists() {
+		return nil, body
+	}
+	var betas []string
+	if betasResult.IsArray() {
+		for _, item := range betasResult.Array() {
+			if s := strings.TrimSpace(item.String()); s != "" {
+				betas = append(betas, s)
+			}
+		}
+	} else if s := strings.TrimSpace(betasResult.String()); s != "" {
+		betas = append(betas, s)
+	}
+	body, _ = sjson.DeleteBytes(body, "betas")
+	return betas, body
+}
+
+// injectThinkingConfig adds thinking configuration based on model name suffix
+func (e *ClaudeExecutor) injectThinkingConfig(modelName string, body []byte) []byte {
+	// Only inject if thinking config is not already present
+	if gjson.GetBytes(body, "thinking").Exists() {
+		return body
+	}
+
+	var budgetTokens int
+	switch {
+	case strings.HasSuffix(modelName, "-thinking-low"):
+		budgetTokens = 1024
+	case strings.HasSuffix(modelName, "-thinking-medium"):
+		budgetTokens = 8192
+	case strings.HasSuffix(modelName, "-thinking-high"):
+		budgetTokens = 24576
+	case strings.HasSuffix(modelName, "-thinking"):
+		// Default thinking without suffix uses medium budget
+		budgetTokens = 8192
+	default:
+		return body
+	}
+
+	body, _ = sjson.SetBytes(body, "thinking.type", "enabled")
+	body, _ = sjson.SetBytes(body, "thinking.budget_tokens", budgetTokens)
+	return body
+}
+
 func (e *ClaudeExecutor) resolveUpstreamModel(alias string, auth *cliproxyauth.Auth) string {
 	if alias == "" {
 		return ""
 	}
+	// Hardcoded mappings for thinking models to actual Claude model names
+	switch alias {
+	case "claude-opus-4-5-thinking", "claude-opus-4-5-thinking-low", "claude-opus-4-5-thinking-medium", "claude-opus-4-5-thinking-high":
+		return "claude-opus-4-5-20251101"
+	case "claude-sonnet-4-5-thinking":
+		return "claude-sonnet-4-5-20250929"
+	}
 	entry := e.resolveClaudeConfig(auth)
 	if entry == nil {
 		return ""
@@ -530,7 +601,7 @@ func decodeResponseBody(body io.ReadCloser, contentEncoding string) (io.ReadClos
 	return body, nil
 }
 
-func applyClaudeHeaders(r *http.Request, auth *cliproxyauth.Auth, apiKey string, stream bool) {
+func applyClaudeHeaders(r *http.Request, auth *cliproxyauth.Auth, apiKey string, stream bool, extraBetas []string) {
 	r.Header.Set("Authorization", "Bearer "+apiKey)
 	r.Header.Set("Content-Type", "application/json")
 
@@ -539,15 +610,30 @@ func applyClaudeHeaders(r *http.Request, auth *cliproxyauth.Auth, apiKey string,
 		ginHeaders = ginCtx.Request.Header
 	}
 
+	baseBetas := "claude-code-20250219,oauth-2025-04-20,interleaved-thinking-2025-05-14,fine-grained-tool-streaming-2025-05-14"
 	if val := strings.TrimSpace(ginHeaders.Get("Anthropic-Beta")); val != "" {
+		baseBetas = val
 		if !strings.Contains(val, "oauth") {
-			val += ",oauth-2025-04-20"
+			baseBetas += ",oauth-2025-04-20"
 		}
-		r.Header.Set("Anthropic-Beta", val)
-	} else {
-		r.Header.Set("Anthropic-Beta", "claude-code-20250219,oauth-2025-04-20,interleaved-thinking-2025-05-14,fine-grained-tool-streaming-2025-05-14")
 	}
 
+	// Merge extra betas from request body
+	if len(extraBetas) > 0 {
+		existingSet := make(map[string]bool)
+		for _, b := range strings.Split(baseBetas, ",") {
+			existingSet[strings.TrimSpace(b)] = true
+		}
+		for _, beta := range extraBetas {
+			beta = strings.TrimSpace(beta)
+			if beta != "" && !existingSet[beta] {
+				baseBetas += "," + beta
+				existingSet[beta] = true
+			}
+		}
+	}
+	r.Header.Set("Anthropic-Beta", baseBetas)
+
 	misc.EnsureHeader(r.Header, ginHeaders, "Anthropic-Version", "2023-06-01")
 	misc.EnsureHeader(r.Header, ginHeaders, "Anthropic-Dangerous-Direct-Browser-Access", "true")
 	misc.EnsureHeader(r.Header, ginHeaders, "X-App", "cli")
@@ -590,3 +676,22 @@ func claudeCreds(a *cliproxyauth.Auth) (apiKey, baseURL string) {
 	}
 	return
 }
+
+func checkSystemInstructions(payload []byte) []byte {
+	system := gjson.GetBytes(payload, "system")
+	claudeCodeInstructions := `[{"type":"text","text":"You are Claude Code, Anthropic's official CLI for Claude.","cache_control":{"type":"ephemeral"}}]`
+	if system.IsArray() {
+		if gjson.GetBytes(payload, "system.0.text").String() != "You are Claude Code, Anthropic's official CLI for Claude." {
+			system.ForEach(func(_, part gjson.Result) bool {
+				if part.Get("type").String() == "text" {
+					claudeCodeInstructions, _ = sjson.SetRaw(claudeCodeInstructions, "-1", part.Raw)
+				}
+				return true
+			})
+			payload, _ = sjson.SetRawBytes(payload, "system", []byte(claudeCodeInstructions))
+		}
+	} else {
+		payload, _ = sjson.SetRawBytes(payload, "system", []byte(claudeCodeInstructions))
+	}
+	return payload
+}

sdk/api/handlers/handlers.go

@@ -4,6 +4,7 @@
 package handlers
 
 import (
+	"bytes"
 	"fmt"
 	"net/http"
 	"strings"
@@ -120,11 +121,11 @@ func (h *BaseAPIHandler) GetContextWithCancel(handler interfaces.APIHandler, c *
 				data := params[0]
 				switch data.(type) {
 				case []byte:
-					c.Set("API_RESPONSE", data.([]byte))
+					appendAPIResponse(c, data.([]byte))
 				case error:
-					c.Set("API_RESPONSE", []byte(data.(error).Error()))
+					appendAPIResponse(c, []byte(data.(error).Error()))
 				case string:
-					c.Set("API_RESPONSE", []byte(data.(string)))
+					appendAPIResponse(c, []byte(data.(string)))
 				case bool:
 				case nil:
 				}
@@ -135,6 +136,28 @@ func (h *BaseAPIHandler) GetContextWithCancel(handler interfaces.APIHandler, c *
 	}
 }
 
+// appendAPIResponse preserves any previously captured API response and appends new data.
+func appendAPIResponse(c *gin.Context, data []byte) {
+	if c == nil || len(data) == 0 {
+		return
+	}
+
+	if existing, exists := c.Get("API_RESPONSE"); exists {
+		if existingBytes, ok := existing.([]byte); ok && len(existingBytes) > 0 {
+			combined := make([]byte, 0, len(existingBytes)+len(data)+1)
+			combined = append(combined, existingBytes...)
+			if existingBytes[len(existingBytes)-1] != '\n' {
+				combined = append(combined, '\n')
+			}
+			combined = append(combined, data...)
+			c.Set("API_RESPONSE", combined)
+			return
+		}
+	}
+
+	c.Set("API_RESPONSE", bytes.Clone(data))
+}
+
 // ExecuteWithAuthManager executes a non-streaming request via the core auth manager.
 // This path is the only supported execution route.
 func (h *BaseAPIHandler) ExecuteWithAuthManager(ctx context.Context, handlerType, modelName string, rawJSON []byte, alt string) ([]byte, *interfaces.ErrorMessage) {
@@ -297,7 +320,7 @@ func (h *BaseAPIHandler) ExecuteStreamWithAuthManager(ctx context.Context, handl
 func (h *BaseAPIHandler) getRequestDetails(modelName string) (providers []string, normalizedModel string, metadata map[string]any, err *interfaces.ErrorMessage) {
 	// Resolve "auto" model to an actual available model first
 	resolvedModelName := util.ResolveAutoModel(modelName)
-	
+
 	providerName, extractedModelName, isDynamic := h.parseDynamicModel(resolvedModelName)
 
 	// First, normalize the model name to handle suffixes like "-thinking-128"