Add OpenAI-compatible audio transcription endpoint

internal/api/server.go

@@ -330,6 +330,7 @@ func (s *Server) setupRoutes() {
 	{
 		v1.GET("/models", s.unifiedModelsHandler(openaiHandlers, claudeCodeHandlers))
 		v1.POST("/chat/completions", openaiHandlers.ChatCompletions)
+		v1.POST("/audio/transcriptions", openaiHandlers.AudioTranscriptions)
 		v1.POST("/completions", openaiHandlers.Completions)
 		v1.POST("/messages", claudeCodeHandlers.ClaudeMessages)
 		v1.POST("/messages/count_tokens", claudeCodeHandlers.ClaudeCountTokens)
@@ -353,6 +354,7 @@ func (s *Server) setupRoutes() {
 			"message": "CLI Proxy API Server",
 			"endpoints": []string{
 				"POST /v1/chat/completions",
+				"POST /v1/audio/transcriptions",
 				"POST /v1/completions",
 				"GET /v1/models",
 			},

internal/api/server_test.go

@@ -208,3 +208,18 @@ func TestDefaultRequestLoggerFactory_UsesResolvedLogDirectory(t *testing.T) {
 		}
 	}
 }
+
+func TestRootEndpointIncludesAudioTranscriptions(t *testing.T) {
+	server := newTestServer(t)
+
+	req := httptest.NewRequest(http.MethodGet, "/", nil)
+	resp := httptest.NewRecorder()
+	server.engine.ServeHTTP(resp, req)
+
+	if resp.Code != http.StatusOK {
+		t.Fatalf("status = %d, want %d; body=%s", resp.Code, http.StatusOK, resp.Body.String())
+	}
+	if body := resp.Body.String(); !strings.Contains(body, "POST /v1/audio/transcriptions") {
+		t.Fatalf("response body missing audio transcription endpoint: %s", body)
+	}
+}

internal/registry/models/models.json

@@ -1371,6 +1371,20 @@
           "xhigh"
         ]
       }
+    },
+    {
+      "id": "gpt-4o-mini-transcribe",
+      "object": "model",
+      "created": 1729728000,
+      "owned_by": "openai",
+      "type": "openai",
+      "display_name": "GPT-4o Mini Transcribe",
+      "version": "gpt-4o-mini-transcribe",
+      "description": "Fast speech-to-text model for audio transcription.",
+      "supported_parameters": [
+        "language",
+        "prompt"
+      ]
     }
   ],
   "codex-team": [
@@ -1623,6 +1637,20 @@
           "xhigh"
         ]
       }
+    },
+    {
+      "id": "gpt-4o-mini-transcribe",
+      "object": "model",
+      "created": 1729728000,
+      "owned_by": "openai",
+      "type": "openai",
+      "display_name": "GPT-4o Mini Transcribe",
+      "version": "gpt-4o-mini-transcribe",
+      "description": "Fast speech-to-text model for audio transcription.",
+      "supported_parameters": [
+        "language",
+        "prompt"
+      ]
     }
   ],
   "codex-plus": [
@@ -1898,6 +1926,20 @@
           "xhigh"
         ]
       }
+    },
+    {
+      "id": "gpt-4o-mini-transcribe",
+      "object": "model",
+      "created": 1729728000,
+      "owned_by": "openai",
+      "type": "openai",
+      "display_name": "GPT-4o Mini Transcribe",
+      "version": "gpt-4o-mini-transcribe",
+      "description": "Fast speech-to-text model for audio transcription.",
+      "supported_parameters": [
+        "language",
+        "prompt"
+      ]
     }
   ],
   "codex-pro": [
@@ -2173,6 +2215,20 @@
           "xhigh"
         ]
       }
+    },
+    {
+      "id": "gpt-4o-mini-transcribe",
+      "object": "model",
+      "created": 1729728000,
+      "owned_by": "openai",
+      "type": "openai",
+      "display_name": "GPT-4o Mini Transcribe",
+      "version": "gpt-4o-mini-transcribe",
+      "description": "Fast speech-to-text model for audio transcription.",
+      "supported_parameters": [
+        "language",
+        "prompt"
+      ]
     }
   ],
   "qwen": [
@@ -2680,4 +2736,4 @@
       "max_completion_tokens": 32768
     }
   ]
-}
+}

internal/runtime/executor/codex_executor.go

@@ -50,14 +50,7 @@ func (e *CodexExecutor) PrepareRequest(req *http.Request, auth *cliproxyauth.Aut
 		return nil
 	}
 	apiKey, _ := codexCreds(auth)
-	if strings.TrimSpace(apiKey) != "" {
-		req.Header.Set("Authorization", "Bearer "+apiKey)
-	}
-	var attrs map[string]string
-	if auth != nil {
-		attrs = auth.Attributes
-	}
-	util.ApplyCustomHeadersFromAttrs(req, attrs)
+	applyCodexPreparedHeaders(req, auth, apiKey, e.cfg)
 	return nil
 }
 
@@ -637,8 +630,28 @@ func (e *CodexExecutor) cacheHelper(ctx context.Context, from sdktranslator.Form
 }
 
 func applyCodexHeaders(r *http.Request, auth *cliproxyauth.Auth, token string, stream bool, cfg *config.Config) {
+	applyCodexPreparedHeaders(r, auth, token, cfg)
+	if r == nil {
+		return
+	}
 	r.Header.Set("Content-Type", "application/json")
-	r.Header.Set("Authorization", "Bearer "+token)
+	if stream {
+		r.Header.Set("Accept", "text/event-stream")
+	} else {
+		r.Header.Set("Accept", "application/json")
+	}
+}
+
+func applyCodexPreparedHeaders(r *http.Request, auth *cliproxyauth.Auth, token string, cfg *config.Config) {
+	if r == nil {
+		return
+	}
+	if r.Header == nil {
+		r.Header = make(http.Header)
+	}
+	if strings.TrimSpace(token) != "" {
+		r.Header.Set("Authorization", "Bearer "+token)
+	}
 
 	var ginHeaders http.Header
 	if ginCtx, ok := r.Context().Value("gin").(*gin.Context); ok && ginCtx != nil && ginCtx.Request != nil {
@@ -649,13 +662,12 @@ func applyCodexHeaders(r *http.Request, auth *cliproxyauth.Auth, token string, s
 	misc.EnsureHeader(r.Header, ginHeaders, "Session_id", uuid.NewString())
 	cfgUserAgent, _ := codexHeaderDefaults(cfg, auth)
 	ensureHeaderWithConfigPrecedence(r.Header, ginHeaders, "User-Agent", cfgUserAgent, codexUserAgent)
-
-	if stream {
-		r.Header.Set("Accept", "text/event-stream")
-	} else {
+	if strings.TrimSpace(r.Header.Get("Accept")) == "" {
 		r.Header.Set("Accept", "application/json")
 	}
-	r.Header.Set("Connection", "Keep-Alive")
+	if strings.TrimSpace(r.Header.Get("Connection")) == "" {
+		r.Header.Set("Connection", "Keep-Alive")
+	}
 
 	isAPIKey := false
 	if auth != nil && auth.Attributes != nil {

internal/runtime/executor/codex_executor_test.go

@@ -0,0 +1,42 @@
+package executor
+
+import (
+	"net/http"
+	"testing"
+
+	cliproxyauth "github.com/router-for-me/CLIProxyAPI/v6/sdk/cliproxy/auth"
+)
+
+func TestCodexPrepareRequestPreservesMultipartContentType(t *testing.T) {
+	req, err := http.NewRequest(http.MethodPost, "https://example.com/backend-api/transcribe", nil)
+	if err != nil {
+		t.Fatalf("NewRequest() error = %v", err)
+	}
+	req.Header.Set("Content-Type", "multipart/form-data; boundary=test-boundary")
+
+	executor := NewCodexExecutor(nil)
+	auth := &cliproxyauth.Auth{
+		Provider: "codex",
+		Metadata: map[string]any{
+			"account_id": "account-123",
+			"email":      "user@example.com",
+		},
+	}
+
+	if err := executor.PrepareRequest(req, auth); err != nil {
+		t.Fatalf("PrepareRequest() error = %v", err)
+	}
+
+	if got := req.Header.Get("Content-Type"); got != "multipart/form-data; boundary=test-boundary" {
+		t.Fatalf("Content-Type = %q, want %q", got, "multipart/form-data; boundary=test-boundary")
+	}
+	if got := req.Header.Get("Chatgpt-Account-Id"); got != "account-123" {
+		t.Fatalf("Chatgpt-Account-Id = %q, want %q", got, "account-123")
+	}
+	if got := req.Header.Get("Originator"); got != "codex_cli_rs" {
+		t.Fatalf("Originator = %q, want %q", got, "codex_cli_rs")
+	}
+	if got := req.Header.Get("Accept"); got != "application/json" {
+		t.Fatalf("Accept = %q, want %q", got, "application/json")
+	}
+}

sdk/api/handlers/handlers.go

@@ -511,6 +511,35 @@ func (h *BaseAPIHandler) ExecuteWithAuthManager(ctx context.Context, handlerType
 	return resp.Payload, FilterUpstreamHeaders(resp.Headers), nil
 }
 
+// ExecuteHTTPRequestWithAuthManager executes a raw HTTP request builder via the core auth manager.
+// This path is intended for provider-native endpoints that do not go through the JSON translator stack.
+func (h *BaseAPIHandler) ExecuteHTTPRequestWithAuthManager(ctx context.Context, modelName string, build coreauth.HTTPRequestBuilder) (*http.Response, *coreauth.Auth, *interfaces.ErrorMessage) {
+	providers, normalizedModel, errMsg := h.getRequestDetails(modelName)
+	if errMsg != nil {
+		return nil, nil, errMsg
+	}
+	reqMeta := requestExecutionMetadata(ctx)
+	reqMeta[coreexecutor.RequestedModelMetadataKey] = normalizedModel
+	opts := coreexecutor.Options{Metadata: reqMeta}
+	resp, auth, err := h.AuthManager.ExecuteHTTPRequest(ctx, providers, normalizedModel, normalizedModel, opts, build)
+	if err != nil {
+		status := http.StatusInternalServerError
+		if se, ok := err.(interface{ StatusCode() int }); ok && se != nil {
+			if code := se.StatusCode(); code > 0 {
+				status = code
+			}
+		}
+		var addon http.Header
+		if he, ok := err.(interface{ Headers() http.Header }); ok && he != nil {
+			if hdr := he.Headers(); hdr != nil {
+				addon = hdr.Clone()
+			}
+		}
+		return nil, nil, &interfaces.ErrorMessage{StatusCode: status, Error: err, Addon: addon}
+	}
+	return resp, auth, nil
+}
+
 // ExecuteCountWithAuthManager executes a non-streaming request via the core auth manager.
 // This path is the only supported execution route.
 func (h *BaseAPIHandler) ExecuteCountWithAuthManager(ctx context.Context, handlerType, modelName string, rawJSON []byte, alt string) ([]byte, http.Header, *interfaces.ErrorMessage) {