diff --git a/artifacts/api-server/src/lib/aiAnalysis.ts b/artifacts/api-server/src/lib/aiAnalysis.ts
index d8daa25..bdd1f64 100644
--- a/artifacts/api-server/src/lib/aiAnalysis.ts
+++ b/artifacts/api-server/src/lib/aiAnalysis.ts
@@ -137,6 +137,12 @@ async function callOpenAiCompatible(
   });
   if (!res.ok) {
     const body = await res.text();
+    if (body.includes("v1/responses")) {
+      throw new Error(
+        `Das Modell "${provider.model}" unterstützt nur /v1/responses, nicht /v1/chat/completions. ` +
+          `Bitte wählen Sie ein Chat-kompatibles Modell (z.\u202fB. gpt-4o, gpt-4-turbo, gpt-3.5-turbo).`,
+      );
+    }
     throw new Error(
       `HTTP ${res.status}: ${redactSecrets(body.slice(0, 300), provider.apiToken)}`,
     );
diff --git a/artifacts/skillguard/src/i18n/locales/de/admin.ts b/artifacts/skillguard/src/i18n/locales/de/admin.ts
index 3b4509c..f81de1a 100644
--- a/artifacts/skillguard/src/i18n/locales/de/admin.ts
+++ b/artifacts/skillguard/src/i18n/locales/de/admin.ts
@@ -16,6 +16,8 @@ export default {
     noneFoundTried: "Keine Modelle gefunden – bitte das Modell manuell eingeben.",
     noneFoundHint:
       "Testen Sie die Verbindung, um verfügbare Modelle automatisch zu laden, oder geben Sie das Modell manuell ein.",
+    responsesOnlyWarning:
+      "Dieses Modell unterstützt nur /v1/responses, nicht /v1/chat/completions. Die KI-Analyse wird fehlschlagen. Bitte wählen Sie ein Chat-kompatibles Modell (z.\u202fB. gpt-4o, gpt-4-turbo).",
   },
   providers: {
     heading: "KI-Provider",
diff --git a/artifacts/skillguard/src/i18n/locales/en/admin.ts b/artifacts/skillguard/src/i18n/locales/en/admin.ts
index ba2edb3..cb898e4 100644
--- a/artifacts/skillguard/src/i18n/locales/en/admin.ts
+++ b/artifacts/skillguard/src/i18n/locales/en/admin.ts
@@ -16,6 +16,8 @@ export default {
     noneFoundTried: "No models found – please enter the model manually.",
     noneFoundHint:
       "Test the connection to load available models automatically, or enter the model manually.",
+    responsesOnlyWarning:
+      "This model only supports /v1/responses, not /v1/chat/completions. AI analysis will fail. Please choose a chat-compatible model (e.g. gpt-4o, gpt-4-turbo).",
   },
   providers: {
     heading: "AI providers",
diff --git a/artifacts/skillguard/src/i18n/locales/es/admin.ts b/artifacts/skillguard/src/i18n/locales/es/admin.ts
index 3dbc7c0..9f4c7cc 100644
--- a/artifacts/skillguard/src/i18n/locales/es/admin.ts
+++ b/artifacts/skillguard/src/i18n/locales/es/admin.ts
@@ -16,6 +16,8 @@ export default {
     noneFoundTried: "No se encontraron modelos: introduzca el modelo manualmente.",
     noneFoundHint:
       "Pruebe la conexión para cargar automáticamente los modelos disponibles, o introduzca el modelo manualmente.",
+    responsesOnlyWarning:
+      "Este modelo solo admite /v1/responses, no /v1/chat/completions. El análisis de IA fallará. Seleccione un modelo compatible con chat (p.\u202fej. gpt-4o, gpt-4-turbo).",
   },
   providers: {
     heading: "Proveedores de IA",
diff --git a/artifacts/skillguard/src/pages/admin.tsx b/artifacts/skillguard/src/pages/admin.tsx
index a955611..2744487 100644
--- a/artifacts/skillguard/src/pages/admin.tsx
+++ b/artifacts/skillguard/src/pages/admin.tsx
@@ -42,14 +42,29 @@ const PROVIDER_PRESETS: Partial<Record<AiProviderApiType, ProviderPreset[]>> = {
   ],
 };
 
-function ModelField({ models, loading, tried, value, onChange }: {
+const RESPONSES_ONLY_RE = /^o\d/i;
+
+function isResponsesOnlyModel(model: string, apiType: string): boolean {
+  return apiType === "openai" && RESPONSES_ONLY_RE.test(model.trim());
+}
+
+function ModelField({ models, loading, tried, value, onChange, apiType }: {
   models: string[];
   loading: boolean;
   tried: boolean;
   value: string;
   onChange: (v: string) => void;
+  apiType: string;
 }) {
   const { t } = useTranslation();
+  const warnResponsesOnly = value && isResponsesOnlyModel(value, apiType);
+
+  const warning = warnResponsesOnly ? (
+    <p className="text-xs text-amber-700 bg-amber-50 border border-amber-200 rounded-md px-2 py-1.5">
+      {t("admin.modelField.responsesOnlyWarning")}
+    </p>
+  ) : null;
+
   if (loading) {
     return (
       <div className="grid gap-2">
@@ -71,6 +86,7 @@ function ModelField({ models, loading, tried, value, onChange }: {
           </SelectContent>
         </Select>
         <p className="text-xs text-muted-foreground">{t("admin.modelField.found", { count: models.length })}</p>
+        {warning}
       </div>
     );
   }
@@ -83,6 +99,7 @@ function ModelField({ models, loading, tried, value, onChange }: {
           ? t("admin.modelField.noneFoundTried")
           : t("admin.modelField.noneFoundHint")}
       </p>
+      {warning}
     </div>
   );
 }
@@ -368,6 +385,7 @@ function ProviderTab() {
                   models={addModels}
                   loading={addModelsLoading}
                   tried={addModelsTried}
+                  apiType={addForm.apiType}
                   value={addForm.model}
                   onChange={(v) => setAddForm(f => ({ ...f, model: v }))}
                 />
@@ -498,6 +516,7 @@ function ProviderTab() {
                         models={editModels}
                         loading={editModelsLoading}
                         tried={editModelsTried}
+                        apiType={editForm.apiType}
                         value={editForm.model}
                         onChange={(v) => setEditForm(f => ({ ...f, model: v }))}
                       />
diff --git a/attached_assets/image_1781645514678.png b/attached_assets/image_1781645514678.png
new file mode 100644
index 0000000..b3641e5
Binary files /dev/null and b/attached_assets/image_1781645514678.png differ