chore: make model optional

Author: lich0821

cmd/desktop/frontend/src/i18n/en.js

@@ -87,9 +87,9 @@ export default {
         modelPlaceholder: 'e.g., claude-sonnet-4-5-20250929',
         modelHelp: 'Optional: Override the model specified in requests',
         modelHelpClaude: 'Optional: Override the model specified in requests',
-        modelHelpOpenAI: 'Required: Specify the OpenAI model to use',
-        modelHelpOpenAI2: 'Required: Specify the OpenAI model (Responses API)',
-        modelHelpGemini: 'Required: Specify the Gemini model to use',
+        modelHelpOpenAI: 'Specify the OpenAI model to use',
+        modelHelpOpenAI2: 'Specify the OpenAI model (Responses API)',
+        modelHelpGemini: 'Specify the Gemini model to use',
         remark: 'Remark',
         remarkHelp: 'Optional: Add a remark for this endpoint',
         cancel: 'Cancel',

cmd/desktop/frontend/src/i18n/zh-CN.js

@@ -87,9 +87,9 @@ export default {
         modelPlaceholder: '例如：claude-sonnet-4-5-20250929',
         modelHelp: '可选：覆盖请求中指定的模型',
         modelHelpClaude: '可选：覆盖请求中指定的模型',
-        modelHelpOpenAI: '必填：指定要使用的 OpenAI 模型',
-        modelHelpOpenAI2: '必填：指定 OpenAI 模型（Responses API）',
-        modelHelpGemini: '必填：指定要使用的 Gemini 模型',
+        modelHelpOpenAI: '指定要使用的 OpenAI 模型',
+        modelHelpOpenAI2: '指定 OpenAI 模型（Responses API）',
+        modelHelpGemini: '指定要使用的 Gemini 模型',
         remark: '备注',
         remarkHelp: '可选：为此端点添加备注说明',
         cancel: '取消',

cmd/desktop/frontend/src/modules/modal.js

@@ -259,10 +259,7 @@ export async function saveEndpoint() {
         return;
     }
 
-    if (transformer !== 'claude' && !model) {
-        showError(t('modal.modelRequired').replace('{transformer}', transformer));
-        return;
-    }
+    // Model is optional: only override when provided.
 
     // Check for duplicate endpoint name
     const configStr = await window.go.main.App.GetConfig();
@@ -322,20 +319,17 @@ export function handleTransformerChange() {
     // Clear fetched models when transformer changes
     clearFetchedModels();
 
+    modelRequired.style.display = 'none';
     if (transformer === 'claude') {
-        modelRequired.style.display = 'none';
         modelInput.placeholder = 'e.g., claude-3-5-sonnet-20241022';
         modelHelpText.textContent = t('modal.modelHelpClaude');
     } else if (transformer === 'openai') {
-        modelRequired.style.display = 'inline';
         modelInput.placeholder = 'e.g., gpt-4-turbo';
         modelHelpText.textContent = t('modal.modelHelpOpenAI');
     } else if (transformer === 'openai2') {
-        modelRequired.style.display = 'inline';
         modelInput.placeholder = 'e.g., gpt-4.1';
         modelHelpText.textContent = t('modal.modelHelpOpenAI2');
     } else if (transformer === 'gemini') {
-        modelRequired.style.display = 'inline';
         modelInput.placeholder = 'e.g., gemini-pro';
         modelHelpText.textContent = t('modal.modelHelpGemini');
     }

internal/proxy/proxy.go

@@ -444,6 +444,9 @@ func (p *Proxy) handleProxy(w http.ResponseWriter, r *http.Request) {
 			cleanedBody = transformedBody
 		}
 		transformedBody = cleanedBody
+		if config.NormalizeAuthMode(endpoint.AuthMode) == config.AuthModeCodexTokenPool {
+			transformedBody = overrideModelInPayload(transformedBody, endpoint.Model)
+		}
 
 		var thinkingEnabled bool
 		if strings.Contains(transformerName, "openai") {

internal/proxy/request.go

@@ -328,6 +328,27 @@ func ensureCodexResponsesPayload(payload []byte) []byte {
 	return updated
 }
 
+func overrideModelInPayload(payload []byte, model string) []byte {
+	if strings.TrimSpace(model) == "" {
+		return payload
+	}
+	trimmed := strings.TrimSpace(string(payload))
+	if trimmed == "" || strings.HasPrefix(trimmed, "[") {
+		return payload
+	}
+
+	var body map[string]interface{}
+	if err := json.Unmarshal(payload, &body); err != nil {
+		return payload
+	}
+	body["model"] = model
+	updated, err := json.Marshal(body)
+	if err != nil {
+		return payload
+	}
+	return updated
+}
+
 // sendRequest sends the HTTP request and returns the response
 func sendRequest(ctx context.Context, proxyReq *http.Request, httpClient *http.Client, cfg *config.Config) (*http.Response, error) {
 	proxyReq = proxyReq.WithContext(ctx)

internal/proxy/request_test.go

@@ -55,6 +55,19 @@ func TestNormalizeTargetPathForBaseURLOnCodexBackend(t *testing.T) {
 	}
 }
 
+func TestOverrideModelInPayload(t *testing.T) {
+	raw := []byte(`{"model":"gpt-5.3-codex","stream":true}`)
+	out := overrideModelInPayload(raw, "gpt-5.2-codex")
+
+	var payload map[string]interface{}
+	if err := json.Unmarshal(out, &payload); err != nil {
+		t.Fatalf("unmarshal failed: %v", err)
+	}
+	if payload["model"] != "gpt-5.2-codex" {
+		t.Fatalf("expected model override to gpt-5.2-codex, got %#v", payload["model"])
+	}
+}
+
 func TestShouldHandleAsStreamingResponseForCodexWithoutContentType(t *testing.T) {
 	endpoint := config.Endpoint{
 		Name:        "TokenPool",