diff --git a/ui/app/(prowler)/lighthouse/_components/chat/lighthouse-v2-chat-page.test.tsx b/ui/app/(prowler)/lighthouse/_components/chat/lighthouse-v2-chat-page.test.tsx
index 5e2bb0cfed..2fc0b96c3b 100644
--- a/ui/app/(prowler)/lighthouse/_components/chat/lighthouse-v2-chat-page.test.tsx
+++ b/ui/app/(prowler)/lighthouse/_components/chat/lighthouse-v2-chat-page.test.tsx
@@ -219,6 +219,25 @@ describe("LighthouseV2ChatPage", () => {
     expect(within(currentModel).getByText("GPT-5.1")).toBeInTheDocument();
   });
 
+  it("defaults to gpt-5.5 when OpenAI has no remembered model", () => {
+    // Given / When
+    renderPage({
+      configurations: [
+        { ...configurations[0], defaultModel: null, connected: true },
+        { ...configurations[1], connected: false },
+      ],
+      modelsByProvider: {
+        openai: [model("gpt-4.1", "GPT-4.1"), model("gpt-5.5", "GPT-5.5")],
+        bedrock: [model("anthropic.claude-4")],
+        "openai-compatible": [model("llama-3.3")],
+      },
+    });
+
+    // Then
+    const currentModel = screen.getByLabelText("Current model: OpenAI GPT-5.5");
+    expect(within(currentModel).getByText("GPT-5.5")).toBeInTheDocument();
+  });
+
   it("uses the AWS onboarding quick prompt instead of the docs prompt", async () => {
     // Given
     const user = userEvent.setup();
diff --git a/ui/app/(prowler)/lighthouse/_components/chat/lighthouse-v2-chat-page.tsx b/ui/app/(prowler)/lighthouse/_components/chat/lighthouse-v2-chat-page.tsx
index 4f5baf6de7..82697ce506 100644
--- a/ui/app/(prowler)/lighthouse/_components/chat/lighthouse-v2-chat-page.tsx
+++ b/ui/app/(prowler)/lighthouse/_components/chat/lighthouse-v2-chat-page.tsx
@@ -516,6 +516,13 @@ const LIGHTHOUSE_V2_PROVIDER_PRIORITY = [
   LIGHTHOUSE_V2_PROVIDER_TYPE.OPENAI_COMPATIBLE,
 ] as const;
 
+// Fallback model per provider when the configuration has no remembered model.
+const LIGHTHOUSE_V2_PREFERRED_DEFAULT_MODEL: Partial<
+  Record<LighthouseV2ProviderType, string>
+> = {
+  [LIGHTHOUSE_V2_PROVIDER_TYPE.OPENAI]: "gpt-5.5",
+};
+
 function resolveInitialModelSelection(
   connectedConfigurations: LighthouseV2Configuration[],
   modelsByProvider: Record<
@@ -535,14 +542,19 @@ function resolveInitialModelSelection(
   for (const configuration of orderedConfigurations) {
     const providerModels = modelsByProvider[configuration.providerType] ?? [];
     if (providerModels.length === 0) continue;
-    // Prefer the provider's remembered model when it is still supported;
-    // otherwise fall back to the first supported model.
+    // Prefer the provider's remembered model when it is still supported, then
+    // the provider's preferred default, then the first supported model.
     const rememberedModel = providerModels.find(
       (model) => model.id === configuration.defaultModel,
     );
+    const preferredModel = providerModels.find(
+      (model) =>
+        model.id ===
+        LIGHTHOUSE_V2_PREFERRED_DEFAULT_MODEL[configuration.providerType],
+    );
     return {
       providerType: configuration.providerType,
-      modelId: (rememberedModel ?? providerModels[0]).id,
+      modelId: (rememberedModel ?? preferredModel ?? providerModels[0]).id,
     };
   }