diff --git a/src/app/llms-full.txt/route.ts b/src/app/llms-full.txt/route.ts
index ea205e21..b87c0b3f 100644
--- a/src/app/llms-full.txt/route.ts
+++ b/src/app/llms-full.txt/route.ts
@@ -1,17 +1,17 @@
-import { redirect } from "next/navigation";
+import { GET as buildLlmContext } from "@/app/api/llm-context/route";
 
 export const runtime = "nodejs";
+export const revalidate = 60;
 
 /**
- * llms-full.txt - the expanded sibling of `/llms.txt` per the emerging
- * llmstxt.org convention. We serve the same content as `/api/llm-context`
- * (full rankings + methodology for every live benchmark) so an LLM that
- * follows the convention finds the rich Markdown view without having to
- * know our internal API path.
+ * llms-full.txt — expanded sibling of /llms.txt per the llmstxt.org
+ * convention. Serves the same body as /api/llm-context, but directly
+ * (200 + text/markdown) instead of via redirect. The previous 307 broke
+ * llms.txt validators and the stricter AI crawlers that do not follow
+ * redirects on non-standard paths (Perplexity, some Bing AI scrapers),
+ * which were marking the URL as no-content and skipping the resource.
  *
- * Internal redirect keeps a single source of truth; the `/llms-full.txt`
- * URL is what crawlers will look up.
+ * The canonical for AI citation is now /llms-full.txt itself, not the
+ * internal /api/llm-context path.
  */
-export function GET() {
-  redirect("/api/llm-context");
-}
+export const GET = buildLlmContext;