components/elements/response.tsx

@@ -18,6 +18,7 @@ import { ResponsePdfDownloadButton } from "@/components/elements/response-pdf-do
 import { buildImageProxyUrl } from "@/lib/image-delivery";
 import { splitMarkdownImages } from "@/lib/canvas/markdown-image-segments";
 import { stableStreamingMarkdown } from "@/lib/canvas/markdown-image-streaming";
+import { getGatewayErrorPageMessage } from "@/lib/request-error";
 import {
   Collapsible,
   CollapsibleContent,
@@ -1028,7 +1029,10 @@ export const Response = ({
 
   const rawText =
     typeof props.children === "string"
-      ? normalizeMarkdownContent(stableStreamingMarkdown(props.children, isMarkdownStreaming))
+      ? normalizeMarkdownContent(stableStreamingMarkdown(
+          getGatewayErrorPageMessage(props.children) || props.children,
+          isMarkdownStreaming
+        ))
       : null;
   const shouldShowPdfDownload =
     Boolean(enablePdfDownload && rawText && hasDownloadableReportContent(rawText));

components/message.tsx

@@ -7,6 +7,7 @@ import type { Vote } from "@/lib/db/schema";
 import type { ChatMessage } from "@/lib/types";
 import { cn, sanitizeText } from "@/lib/utils";
 import { extractImageUrlsFromText } from "@/lib/image-url";
+import { getGatewayErrorPageMessage } from "@/lib/request-error";
 import { useDataStream } from "./data-stream-provider";
 import { DocumentToolResult } from "./document";
 import { DocumentPreview } from "./document-preview";
@@ -521,7 +522,7 @@ const PurePreviewMessage = ({
                 const displayModelName = extractedModelName || inferredFallbackModelName;
                 const filteredText =
                   message.role === "assistant"
-                    ? appendImageDoneSummary(contentWithoutBadge)
+                    ? getGatewayErrorPageMessage(contentWithoutBadge) || appendImageDoneSummary(contentWithoutBadge)
                     : contentWithoutBadge;
                 const useAssistantTypewriter =
                   message.role === "assistant" &&

lib/ai/providers/vip98k.ts

@@ -48,6 +48,7 @@ import {
   VIP98K_REFERENCE_STAGING_KEY_PREFIX,
 } from "@/lib/ai/vip98k-reference-staging";
 import { createLogger } from "@/lib/observability/logger";
+import { getUserFacingImageGenerationErrorMessage } from "@/lib/request-error";
 import { createDeferredUploadedImage } from "@/lib/storage/image-archive";
 import {
   IMAGE_UPLOAD_MAX_BYTES,
@@ -1188,9 +1189,9 @@ async function submitTask(params: {
       continue;
     }
     if ((response.status === 409 && /idempotency\s+key.*(?:still\s+in\s+progress|processing)/i.test(errorMessage))
-      || [502, 504].includes(response.status)
+      || [502, 504, 524].includes(response.status)
       || /pending\s+safe\s+reconciliation|request\s+outcome\s+(?:is\s+)?unknown|final\s+outcome\s+could\s+not\s+be\s+confirmed/i.test(errorMessage))
-      throw new ProviderTaskUncertainError(`供应商提交结果未确认，可能仍在处理，请勿重复生成：${errorMessage} (HTTP ${response.status})`);
+      throw new ProviderTaskUncertainError(`供应商提交结果未确认，可能仍在处理，请勿重复生成：${getUserFacingImageGenerationErrorMessage({ message: errorMessage, status: response.status })} (HTTP ${response.status})`);
     throw new Error(`${errorMessage} (HTTP ${response.status})`);
   }
 

lib/ai/runtime/model-health.ts

@@ -12,7 +12,7 @@ import {
   isMediaInputFailure,
   isNonProviderRequestStatus,
 } from "@/lib/ai/runtime/request-failure-classification";
-import { isProviderBalanceError, isRateLimitError } from "@/lib/request-error";
+import { isGatewayResponseTimeout, isProviderBalanceError, isRateLimitError } from "@/lib/request-error";
 
 export type AiModelHealthEventType =
   | "success"
@@ -343,6 +343,7 @@ export function classifyAiModelHealthFailure(error: unknown): {
 }
 
 export function shouldFailoverAiModelError(error: unknown) {
+  if (isGatewayResponseTimeout(error)) return false;
   const classification = classifyAiModelHealthFailure(error);
   const normalized = getErrorMessage(error).toLowerCase();
   const imageProviderTaskFailed =

lib/request-error.ts

@@ -175,6 +175,28 @@ function getRequestErrorText(error: unknown) {
   return candidates.join(" ").slice(0, 4000);
 }
 
+export function getGatewayErrorPageMessage(content: string): string | null {
+  if (!/^\s*(?:<!doctype\s+html\b|<html\b)/i.test(content)) return null;
+  if (!/\bcf-error-details\b/i.test(content)) return null;
+  if (/\b524\b/.test(content)) {
+    return "模型服务响应超时（HTTP 524），生成结果尚未确认。请先查看历史记录，避免重复提交。";
+  }
+  return "模型服务暂时不可用，请稍后再试或联系管理员。";
+}
+
+export function isGatewayResponseTimeout(error: unknown): boolean {
+  return getRequestErrorStatus(error) === 524 ||
+    /\bHTTP\s*524\b|\b524\s*:\s*A timeout occurred\b|Error code\s*524\b/i.test(getRequestErrorText(error));
+}
+
+function getPublicGatewayErrorMessage(error: unknown): string | null {
+  if (isGatewayResponseTimeout(error)) {
+    return "模型服务响应超时（HTTP 524），生成结果尚未确认。请先查看历史记录，避免重复提交。";
+  }
+  const message = getRequestErrorMessage(error);
+  return message ? getGatewayErrorPageMessage(message) : null;
+}
+
 export function isProviderBalanceError(error: unknown): boolean {
   const message = getRequestErrorText(error);
   return PROVIDER_BALANCE_PATTERNS.some((pattern) => pattern.test(message));
@@ -203,6 +225,8 @@ export function getUserFacingImageGenerationErrorMessage(
   error: unknown,
   fallbackMessage = "图片生成失败，请稍后重试",
 ) {
+  const gatewayMessage = getPublicGatewayErrorMessage(error);
+  if (gatewayMessage) return gatewayMessage;
   if (isProviderBalanceError(error)) {
     return "图片服务商线路余额或额度不足，请切换模型或联系管理员；与您的积分余额无关。";
   }
@@ -227,6 +251,8 @@ export function getUserFacingRequestErrorMessage(
     rateLimitMessage?: string;
   }
 ) {
+  const gatewayMessage = getPublicGatewayErrorMessage(error);
+  if (gatewayMessage) return gatewayMessage;
   if (isProviderBalanceError(error)) {
     return "模型服务商线路余额或额度不足，请切换模型或联系管理员。";
   }

public/particle-scatter-effect.html

@@ -4,7 +4,6 @@
     <meta charset="UTF-8">
     <meta name="viewport" content="width=device-width, initial-scale=1.0">
     <title>Neural Particle - Scatter Effect</title>
-    <script src="https://cdn.jsdelivr.net/npm/three@0.160.0/build/three.min.js"></script>
     <style>
         * { margin: 0; padding: 0; box-sizing: border-box; }
         body { overflow: hidden; background-color: #000; }
@@ -14,7 +13,8 @@
 </head>
 <body>
     <div id="particle-container"></div>
-    <script>
+    <script type="module">
+        import * as THREE from "/three/0.160.0/three.module.min.js";
         const CONFIG = {
             effect: "scatter",
             effectMode: 1,

public/three/0.160.0/LICENSE

@@ -0,0 +1,21 @@
+The MIT License
+
+Copyright © 2010-2023 three.js authors
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in
+all copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+THE SOFTWARE.

tests/routes/image-provider-error-classification.test.ts

@@ -2,6 +2,8 @@ import { expect, test } from "@playwright/test";
 import {
   getUserFacingImageGenerationErrorMessage,
   getUserFacingRequestErrorMessage,
+  getGatewayErrorPageMessage,
+  isGatewayResponseTimeout,
   isProviderBalanceError,
   isRateLimitError,
 } from "@/lib/request-error";
@@ -16,6 +18,49 @@ test("a bare HTTP 429 is rate limiting, not insufficient balance", () => {
   expect(getUserFacingRequestErrorMessage(error)).toContain("请求过于频繁");
 });
 
+const gatewayHtml = '<!DOCTYPE html><html><head><title>vip98k.com | 524: A timeout occurred</title></head><body><div id="cf-error-details">Error code 524</div><script>window.gatewayDebug = true</script></body></html> (HTTP 524)';
+
+test("gateway timeout pages become a concise unconfirmed-result message", () => {
+  for (const error of [new Error(gatewayHtml), { status: 524, message: "Gateway failed" }, { error: { message: "Request failed (HTTP 524)" } }]) {
+    expect(isGatewayResponseTimeout(error)).toBe(true);
+    for (const text of [getUserFacingImageGenerationErrorMessage(error), getUserFacingRequestErrorMessage(error)]) {
+      expect(text).toContain("HTTP 524");
+      expect(text).toContain("结果尚未确认");
+      expect(text).not.toMatch(/<html|<script|cf-error-details/);
+    }
+  }
+  expect(getGatewayErrorPageMessage(gatewayHtml)).toContain("避免重复提交");
+});
+
+test("ordinary HTML documents and fenced examples retain their content", () => {
+  for (const text of ['<!DOCTYPE html><html><title>My page</title><p>Hosted on Cloudflare</p></html>', `\`\`\`html\n${gatewayHtml}\n\`\`\``, "Cloudflare 文档示例", "普通图片错误"]) {
+    expect(getGatewayErrorPageMessage(text)).toBeNull();
+  }
+});
+
+test("HTTP 524 does not start another model while the original result is unconfirmed", () => {
+  const { health } = healthHarness();
+  expect(health.shouldFailoverAiModelError(new Error(gatewayHtml))).toBe(false);
+  expect(health.shouldFailoverAiModelError({ status: 524, message: "Gateway timeout" })).toBe(false);
+  expect(health.shouldFailoverAiModelError(new Error("Service unavailable (HTTP 503)"))).toBe(true);
+});
+
+test("the provider submits once and keeps an HTTP 524 response unconfirmed", async () => {
+  let submissions = 0;
+  const provider = serverLoader({
+    "@/lib/db/queries": { getSystemSetting: async () => null },
+    "@/lib/storage/image-archive": {},
+    "@/lib/storage/service": {},
+    fetch: async () => {
+      submissions++;
+      return new Response(gatewayHtml, { status: 524, headers: { "content-type": "text/html" } });
+    },
+  })("lib/ai/providers/vip98k.ts");
+  const model = provider.createVip98k({ apiKey: "test", baseURL: "https://provider.test" }).chat("gpt-image-2.5");
+  await expect(model.doGenerate({ messages: [{ role: "user", content: "A desk" }] })).rejects.toThrow(/结果未确认.*HTTP 524/);
+  expect(submissions).toBe(1);
+});
+
 test("an explicit provider balance error takes precedence over HTTP 429", () => {
   const error = Object.assign(new Error("余额不足，请点击充值页面续费（HTTP 429）"), { status: 429 });
   expect(isProviderBalanceError(error)).toBe(true);
@@ -62,8 +107,10 @@ function healthHarness() {
   let model: any = { id: "test", modelId: "test", isEnabled: true, excludeFromFailover: true,
     excludeFromHealthBan: true, healthStatus: "healthy", healthConsecutiveFailures: 0 };
   const events: any[] = [];
-  const query = (rows: () => any[]) => ({ from() { return this; }, where() { return this; },
-    limit: async () => rows(), then: (resolve: any) => Promise.resolve(rows()).then(resolve), returning: async () => rows() });
+  const query = (rows: () => any[]) => Object.assign(Promise.resolve().then(rows), {
+    from() { return this; }, where() { return this; },
+    limit: async () => rows(), returning: async () => rows(),
+  });
   const db = {
     select: () => query(() => [model]),
     insert: () => ({ values: async (value: any) => { events.push(value); } }),