sidecar: default model gpt-5.4-mini → gpt-5.6-luna + codex-spark in search only

- Default sidecar model changed from gpt-5.4-mini to gpt-5.6-luna
  (web-search, vision, management API fallbacks, GUI defaults)
- Dashboard: split sidecar model lists — search includes
  gpt-5.3-codex-spark, vision excludes it
- Startup migration: auto-migrate existing gpt-5.4-mini sidecar configs
  to luna when current time is past KST 2026-06-10 06:00
- Tests updated for new default model

Verified: 1759/1759 tests pass, tsc clean.

Author: lidge-jun

gui/src/pages/Dashboard.tsx

@@ -49,7 +49,8 @@ interface UpdateJob {
   restarted?: boolean;
 }
 
-const SIDECAR_MODELS = ["gpt-5.4-mini", "gpt-5.4", "gpt-5.5", "gpt-5.6-sol", "gpt-5.6-terra", "gpt-5.6-luna"];
+const SEARCH_SIDECAR_MODELS = ["gpt-5.6-luna", "gpt-5.4-mini", "gpt-5.4", "gpt-5.5", "gpt-5.3-codex-spark", "gpt-5.6-sol", "gpt-5.6-terra"];
+const VISION_SIDECAR_MODELS = ["gpt-5.6-luna", "gpt-5.4-mini", "gpt-5.4", "gpt-5.5", "gpt-5.6-sol", "gpt-5.6-terra"];
 const REASONING_LEVELS = ["low", "medium", "high"];
 
 function defaultUpdateChannel(version: string | undefined): UpdateChannel {
@@ -440,8 +441,8 @@ export default function Dashboard({ apiBase }: { apiBase: string }) {
           </div>
           <div className="setting-controls" style={{ display: "flex", gap: 8 }}>
             <Select
-              value={sidecar?.webSearch.model ?? "gpt-5.4-mini"}
-              options={SIDECAR_MODELS.map(m => ({ value: m, label: m }))}
+              value={sidecar?.webSearch.model ?? "gpt-5.6-luna"}
+              options={SEARCH_SIDECAR_MODELS.map(m => ({ value: m, label: m }))}
               onChange={v => saveSidecar({ webSearch: { model: v, reasoning: sidecar!.webSearch.reasoning } })}
               disabled={!sidecar || sidecarSaving}
               label={t("dash.searchModel")}
@@ -464,8 +465,8 @@ export default function Dashboard({ apiBase }: { apiBase: string }) {
             <div className="muted setting-hint">{t("dash.visionModelHint")}</div>
           </div>
           <Select
-            value={sidecar?.vision.model ?? "gpt-5.4-mini"}
-            options={SIDECAR_MODELS.map(m => ({ value: m, label: m }))}
+            value={sidecar?.vision.model ?? "gpt-5.6-luna"}
+            options={VISION_SIDECAR_MODELS.map(m => ({ value: m, label: m }))}
             onChange={v => saveSidecar({ vision: { model: v } })}
             disabled={!sidecar || sidecarSaving}
             label={t("dash.visionModel")}

src/server/index.ts

@@ -158,6 +158,24 @@ export function startServer(port?: number) {
     config.subagentModels = [...DEFAULT_SUBAGENT_MODELS];
     saveConfig(config);
   }
+  // Sidecar model migration (KST 2026-06-10 06:00 = UTC 2026-06-09 21:00): auto-migrate the old
+  // gpt-5.4-mini default to gpt-5.6-luna for both search and vision sidecars. Only touches configs
+  // still on the old default — explicit user choices are preserved.
+  {
+    const SIDECAR_MIGRATION_CUTOFF = Date.UTC(2026, 5, 9, 21, 0); // June 9 21:00 UTC = KST June 10 06:00
+    if (Date.now() >= SIDECAR_MIGRATION_CUTOFF) {
+      let migrated = false;
+      if (config.webSearchSidecar?.model === "gpt-5.4-mini") {
+        config.webSearchSidecar = { ...config.webSearchSidecar, model: "gpt-5.6-luna" };
+        migrated = true;
+      }
+      if (config.visionSidecar?.model === "gpt-5.4-mini") {
+        config.visionSidecar = { ...config.visionSidecar, model: "gpt-5.6-luna" };
+        migrated = true;
+      }
+      if (migrated) saveConfig(config);
+    }
+  }
   invalidateCodexModelsCache();
 
   const listenPort = port ?? config.port ?? 10100;

src/server/management-api.ts

@@ -138,8 +138,8 @@ export async function handleManagementAPI(req: Request, url: URL, config: OcxCon
     const ws = config.webSearchSidecar ?? {};
     const vs = config.visionSidecar ?? {};
     return jsonResponse({
-      webSearch: { model: ws.model ?? "gpt-5.4-mini", reasoning: ws.reasoning ?? "low" },
-      vision: { model: vs.model ?? "gpt-5.4-mini" },
+      webSearch: { model: ws.model ?? "gpt-5.6-luna", reasoning: ws.reasoning ?? "low" },
+      vision: { model: vs.model ?? "gpt-5.6-luna" },
     });
   }
 
@@ -160,8 +160,8 @@ export async function handleManagementAPI(req: Request, url: URL, config: OcxCon
     const vs = config.visionSidecar ?? {};
     return jsonResponse({
       ok: true,
-      webSearch: { model: ws.model ?? "gpt-5.4-mini", reasoning: ws.reasoning ?? "low" },
-      vision: { model: vs.model ?? "gpt-5.4-mini" },
+      webSearch: { model: ws.model ?? "gpt-5.6-luna", reasoning: ws.reasoning ?? "low" },
+      vision: { model: vs.model ?? "gpt-5.6-luna" },
     });
   }
 

src/web-search/index.ts

@@ -6,7 +6,7 @@ import type { CodexAuthContext } from "../codex/auth-context";
 export { runWithWebSearch } from "./loop";
 export { buildWebSearchTool, extractHostedWebSearch, WEB_SEARCH_TOOL_NAME } from "./synthetic-tool";
 
-const DEFAULT_SIDECAR_MODEL = "gpt-5.4-mini";
+const DEFAULT_SIDECAR_MODEL = "gpt-5.6-luna";
 // "low" is the lightest effort the ChatGPT backend allows with web_search ("minimal" is rejected:
 // "tools cannot be used with reasoning.effort 'minimal'") — keeps the sidecar fast/cheap.
 const DEFAULT_SIDECAR_REASONING = "low";

tests/e2e-style/phase100-native-parity.test.ts

@@ -112,7 +112,7 @@ describe("Phase 100 Codex-native parity smoke", () => {
     expect(searchPlan).toMatchObject({
       forwardProvider,
       settings: {
-        model: "gpt-5.4-mini",
+        model: "gpt-5.6-luna",
         describeImages: true,
       },
     });

tests/web-search.test.ts

@@ -65,7 +65,7 @@ describe("web-search sidecar planning", () => {
     expect(plan).toBeDefined();
     expect(plan?.forwardProvider).toBe(forwardProvider);
     expect(plan?.hostedTool).toEqual(parsed._webSearch);
-    expect(plan?.settings.model).toBe("gpt-5.4-mini");
+    expect(plan?.settings.model).toBe("gpt-5.6-luna");
   });
 
   test("planWebSearch activates for pool-selected headers even when raw inbound auth would be main", () => {