qwenproxy-cli 1.0.32 → 1.1.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "qwenproxy-cli",
-  "version": "1.0.32",
+  "version": "1.1.0",
   "description": "High-performance OpenAI & Anthropic compatible API gateway for Qwen with multi-account rotation, interactive TUI, and resilient tool calling.",
   "main": "src/index.ts",
   "bin": {
@@ -11,7 +11,7 @@
     "src",
     "!src/tests",
     "!src/benchmarks",
-    "README.md",
+    "README*.md",
     "LICENSE"
   ],
   "scripts": {
@@ -26,8 +26,8 @@
     "reset": "npx tsx src/reset-cooldowns.ts",
     "purge": "npx tsx src/delete-chats.ts",
     "test": "npm run test:mock && npm run test:live",
-    "test:mock": "tsx --test --test-concurrency=1 --env-file=.env.test src/tests/account-concurrency.test.ts src/tests/concurrency-latest-wins.test.ts src/tests/mutex.test.ts src/tests/context-truncation.test.ts src/tests/streaming-thinking-summary.test.ts src/tests/tool-call-guard.test.ts src/tests/utils-logger-full.test.ts src/tests/logger-toolcall-errors.test.ts src/tests/logger-toolcall-disabled.test.ts src/tests/utils-json-full.test.ts src/tests/utils-session-guard-full.test.ts src/tests/core-logic-full.test.ts src/tests/core-logic-accounts.test.ts src/tests/chat-validation-full.test.ts src/tests/chat-logic-full.test.ts src/tests/account-stickiness.test.ts src/tests/accounts-security.test.ts src/tests/advanced.test.ts src/tests/auth-playwright.test.ts src/tests/captcha-solver.test.ts src/tests/cache-scan-regex.test.ts src/tests/chat-cleanup.test.ts src/tests/chat-cleanup-auto.test.ts src/tests/chromium-args.test.ts src/tests/config-current.test.ts src/tests/context-meter.test.ts src/tests/concurrency.test.ts src/tests/delta.test.ts src/tests/fingerprint.test.ts src/tests/human-behavior.test.ts src/tests/index.test.ts src/tests/json-recovery.test.ts src/tests/jsonFix.test.ts src/tests/memory-cache.test.ts src/tests/model-alias.test.ts src/tests/model-registry-thinking.test.ts src/tests/models-endpoint.test.ts src/tests/multimodal.test.ts src/tests/oss-upload.test.ts src/tests/parser.test.ts src/tests/personalization-flow.test.ts src/tests/personalization-required.test.ts src/tests/retry-policy.test.ts src/tests/rotation.test.ts src/tests/server-lifecycle.test.ts src/tests/session-keeper.test.ts src/tests/session-state.test.ts src/tests/single-account-mode.test.ts src/tests/stream-retry.test.ts src/tests/tool-calls-endpoint.test.ts src/tests/tool-instructions-cache.test.ts src/tests/session-id.test.ts src/tests/watchdog-ram.test.ts src/tests/tool-recovery-t1t2.test.ts src/tests/abort-silent-t4.test.ts src/tests/t3-close-tag-escape.test.ts src/tests/t5-tool-cap.test.ts src/tests/logger-redaction.test.ts src/tests/stall-timeout-retryable.test.ts src/tests/t6-missing-quote-array-close.test.ts src/tests/watchdog-rss.test.ts src/tests/logical-thread-batch.test.ts src/tests/chat-pool-style.test.ts src/tests/queue-wait-cap.test.ts src/tests/context-eviction-order.test.ts src/tests/chat-in-progress-settle.test.ts src/tests/stream-lock-leak.test.ts src/tests/stream-semaphore.test.ts src/tests/waf-isolation.test.ts src/tests/legacy-completions.test.ts src/tests/response-format.test.ts src/tests/cors-headers.test.ts src/tests/stream-lock-terminal-cancel.test.ts src/tests/parser-truncated-tool-call.test.ts src/tests/metrics-prometheus.test.ts src/tests/context-death-cleanup.test.ts src/tests/personalization-direct-fetch.test.ts src/tests/completion-headers-parity.test.ts src/tests/parallel-escape-repro.test.ts src/tests/chat-in-progress-scale-cost.test.ts src/tests/temp-chat-mode.test.ts src/tests/profile-reset.test.ts src/tests/error-classification.test.ts src/tests/quota-cooldown.test.ts src/tests/account-ready-gate.test.ts src/tests/tool-cap-notice.test.ts src/tests/interaction-flows.test.ts src/tests/headers-gate-cooldown-fallback.test.ts src/tests/unwrapped-tool-calls.test.ts src/tests/anthropic.test.ts src/tests/sync-clients.test.ts src/tests/sync-top10-clients.test.ts src/tests/cli.test.ts src/tests/paths.test.ts src/tests/tool-manifest.test.ts src/tests/tui.test.ts src/tests/tui-settings.test.ts src/tests/personalization-deadline.test.ts src/tests/update-cli.test.ts",
-    "test:mock:fast": "tsx --test --test-concurrency=2 --env-file=.env.test src/tests/account-concurrency.test.ts src/tests/concurrency-latest-wins.test.ts src/tests/mutex.test.ts src/tests/context-truncation.test.ts src/tests/streaming-thinking-summary.test.ts src/tests/tool-call-guard.test.ts src/tests/utils-logger-full.test.ts src/tests/logger-toolcall-errors.test.ts src/tests/logger-toolcall-disabled.test.ts src/tests/utils-json-full.test.ts src/tests/utils-session-guard-full.test.ts src/tests/core-logic-full.test.ts src/tests/core-logic-accounts.test.ts src/tests/chat-validation-full.test.ts src/tests/chat-logic-full.test.ts src/tests/account-stickiness.test.ts src/tests/accounts-security.test.ts src/tests/advanced.test.ts src/tests/auth-playwright.test.ts src/tests/captcha-solver.test.ts src/tests/cache-scan-regex.test.ts src/tests/chat-cleanup.test.ts src/tests/chat-cleanup-auto.test.ts src/tests/chromium-args.test.ts src/tests/config-current.test.ts src/tests/context-meter.test.ts src/tests/concurrency.test.ts src/tests/delta.test.ts src/tests/fingerprint.test.ts src/tests/human-behavior.test.ts src/tests/index.test.ts src/tests/json-recovery.test.ts src/tests/jsonFix.test.ts src/tests/memory-cache.test.ts src/tests/model-alias.test.ts src/tests/model-registry-thinking.test.ts src/tests/models-endpoint.test.ts src/tests/multimodal.test.ts src/tests/oss-upload.test.ts src/tests/parser.test.ts src/tests/personalization-flow.test.ts src/tests/personalization-required.test.ts src/tests/retry-policy.test.ts src/tests/rotation.test.ts src/tests/server-lifecycle.test.ts src/tests/session-keeper.test.ts src/tests/session-state.test.ts src/tests/single-account-mode.test.ts src/tests/stream-retry.test.ts src/tests/tool-calls-endpoint.test.ts src/tests/tool-instructions-cache.test.ts src/tests/session-id.test.ts src/tests/watchdog-ram.test.ts src/tests/tool-recovery-t1t2.test.ts src/tests/abort-silent-t4.test.ts src/tests/t3-close-tag-escape.test.ts src/tests/t5-tool-cap.test.ts src/tests/logger-redaction.test.ts src/tests/stall-timeout-retryable.test.ts src/tests/t6-missing-quote-array-close.test.ts src/tests/watchdog-rss.test.ts src/tests/logical-thread-batch.test.ts src/tests/chat-pool-style.test.ts src/tests/queue-wait-cap.test.ts src/tests/context-eviction-order.test.ts src/tests/chat-in-progress-settle.test.ts src/tests/stream-lock-leak.test.ts src/tests/stream-semaphore.test.ts src/tests/waf-isolation.test.ts src/tests/legacy-completions.test.ts src/tests/response-format.test.ts src/tests/cors-headers.test.ts src/tests/stream-lock-terminal-cancel.test.ts src/tests/parser-truncated-tool-call.test.ts src/tests/metrics-prometheus.test.ts src/tests/context-death-cleanup.test.ts src/tests/personalization-direct-fetch.test.ts src/tests/completion-headers-parity.test.ts src/tests/parallel-escape-repro.test.ts src/tests/chat-in-progress-scale-cost.test.ts src/tests/temp-chat-mode.test.ts src/tests/profile-reset.test.ts src/tests/error-classification.test.ts src/tests/quota-cooldown.test.ts src/tests/account-ready-gate.test.ts src/tests/tool-cap-notice.test.ts src/tests/interaction-flows.test.ts src/tests/headers-gate-cooldown-fallback.test.ts src/tests/unwrapped-tool-calls.test.ts src/tests/anthropic.test.ts src/tests/sync-clients.test.ts src/tests/sync-top10-clients.test.ts src/tests/cli.test.ts src/tests/paths.test.ts src/tests/tool-manifest.test.ts src/tests/tui.test.ts src/tests/tui-settings.test.ts src/tests/personalization-deadline.test.ts src/tests/update-cli.test.ts",
+    "test:mock": "tsx --test --test-concurrency=1 --env-file=.env.test src/tests/account-concurrency.test.ts src/tests/concurrency-latest-wins.test.ts src/tests/mutex.test.ts src/tests/context-truncation.test.ts src/tests/streaming-thinking-summary.test.ts src/tests/tool-call-guard.test.ts src/tests/utils-logger-full.test.ts src/tests/logger-toolcall-errors.test.ts src/tests/logger-toolcall-disabled.test.ts src/tests/utils-json-full.test.ts src/tests/utils-session-guard-full.test.ts src/tests/core-logic-full.test.ts src/tests/core-logic-accounts.test.ts src/tests/chat-validation-full.test.ts src/tests/chat-logic-full.test.ts src/tests/account-stickiness.test.ts src/tests/accounts-security.test.ts src/tests/advanced.test.ts src/tests/auth-playwright.test.ts src/tests/captcha-solver.test.ts src/tests/cache-scan-regex.test.ts src/tests/chat-cleanup.test.ts src/tests/chat-cleanup-auto.test.ts src/tests/chromium-args.test.ts src/tests/config-current.test.ts src/tests/context-meter.test.ts src/tests/concurrency.test.ts src/tests/delta.test.ts src/tests/fingerprint.test.ts src/tests/human-behavior.test.ts src/tests/index.test.ts src/tests/json-recovery.test.ts src/tests/jsonFix.test.ts src/tests/memory-cache.test.ts src/tests/model-alias.test.ts src/tests/model-registry-thinking.test.ts src/tests/models-endpoint.test.ts src/tests/multimodal.test.ts src/tests/oss-upload.test.ts src/tests/parser.test.ts src/tests/personalization-flow.test.ts src/tests/personalization-required.test.ts src/tests/retry-policy.test.ts src/tests/rotation.test.ts src/tests/server-lifecycle.test.ts src/tests/session-keeper.test.ts src/tests/session-state.test.ts src/tests/single-account-mode.test.ts src/tests/stream-retry.test.ts src/tests/tool-calls-endpoint.test.ts src/tests/tool-instructions-cache.test.ts src/tests/session-id.test.ts src/tests/watchdog-ram.test.ts src/tests/tool-recovery-t1t2.test.ts src/tests/abort-silent-t4.test.ts src/tests/t3-close-tag-escape.test.ts src/tests/t5-tool-cap.test.ts src/tests/logger-redaction.test.ts src/tests/stall-timeout-retryable.test.ts src/tests/t6-missing-quote-array-close.test.ts src/tests/watchdog-rss.test.ts src/tests/logical-thread-batch.test.ts src/tests/chat-pool-style.test.ts src/tests/queue-wait-cap.test.ts src/tests/context-eviction-order.test.ts src/tests/chat-in-progress-settle.test.ts src/tests/stream-lock-leak.test.ts src/tests/stream-semaphore.test.ts src/tests/waf-isolation.test.ts src/tests/legacy-completions.test.ts src/tests/response-format.test.ts src/tests/cors-headers.test.ts src/tests/stream-lock-terminal-cancel.test.ts src/tests/parser-truncated-tool-call.test.ts src/tests/metrics-prometheus.test.ts src/tests/context-death-cleanup.test.ts src/tests/personalization-direct-fetch.test.ts src/tests/completion-headers-parity.test.ts src/tests/parallel-escape-repro.test.ts src/tests/chat-in-progress-scale-cost.test.ts src/tests/temp-chat-mode.test.ts src/tests/profile-reset.test.ts src/tests/error-classification.test.ts src/tests/quota-cooldown.test.ts src/tests/account-ready-gate.test.ts src/tests/tool-cap-notice.test.ts src/tests/interaction-flows.test.ts src/tests/headers-gate-cooldown-fallback.test.ts src/tests/unwrapped-tool-calls.test.ts src/tests/anthropic.test.ts src/tests/sync-clients.test.ts src/tests/sync-top10-clients.test.ts src/tests/cli.test.ts src/tests/paths.test.ts src/tests/tool-manifest.test.ts src/tests/tui.test.ts src/tests/tui-settings.test.ts src/tests/personalization-deadline.test.ts src/tests/update-cli.test.ts src/tests/batch-accounts.test.ts",
+    "test:mock:fast": "tsx --test --test-concurrency=2 --env-file=.env.test src/tests/account-concurrency.test.ts src/tests/concurrency-latest-wins.test.ts src/tests/mutex.test.ts src/tests/context-truncation.test.ts src/tests/streaming-thinking-summary.test.ts src/tests/tool-call-guard.test.ts src/tests/utils-logger-full.test.ts src/tests/logger-toolcall-errors.test.ts src/tests/logger-toolcall-disabled.test.ts src/tests/utils-json-full.test.ts src/tests/utils-session-guard-full.test.ts src/tests/core-logic-full.test.ts src/tests/core-logic-accounts.test.ts src/tests/chat-validation-full.test.ts src/tests/chat-logic-full.test.ts src/tests/account-stickiness.test.ts src/tests/accounts-security.test.ts src/tests/advanced.test.ts src/tests/auth-playwright.test.ts src/tests/captcha-solver.test.ts src/tests/cache-scan-regex.test.ts src/tests/chat-cleanup.test.ts src/tests/chat-cleanup-auto.test.ts src/tests/chromium-args.test.ts src/tests/config-current.test.ts src/tests/context-meter.test.ts src/tests/concurrency.test.ts src/tests/delta.test.ts src/tests/fingerprint.test.ts src/tests/human-behavior.test.ts src/tests/index.test.ts src/tests/json-recovery.test.ts src/tests/jsonFix.test.ts src/tests/memory-cache.test.ts src/tests/model-alias.test.ts src/tests/model-registry-thinking.test.ts src/tests/models-endpoint.test.ts src/tests/multimodal.test.ts src/tests/oss-upload.test.ts src/tests/parser.test.ts src/tests/personalization-flow.test.ts src/tests/personalization-required.test.ts src/tests/retry-policy.test.ts src/tests/rotation.test.ts src/tests/server-lifecycle.test.ts src/tests/session-keeper.test.ts src/tests/session-state.test.ts src/tests/single-account-mode.test.ts src/tests/stream-retry.test.ts src/tests/tool-calls-endpoint.test.ts src/tests/tool-instructions-cache.test.ts src/tests/session-id.test.ts src/tests/watchdog-ram.test.ts src/tests/tool-recovery-t1t2.test.ts src/tests/abort-silent-t4.test.ts src/tests/t3-close-tag-escape.test.ts src/tests/t5-tool-cap.test.ts src/tests/logger-redaction.test.ts src/tests/stall-timeout-retryable.test.ts src/tests/t6-missing-quote-array-close.test.ts src/tests/watchdog-rss.test.ts src/tests/logical-thread-batch.test.ts src/tests/chat-pool-style.test.ts src/tests/queue-wait-cap.test.ts src/tests/context-eviction-order.test.ts src/tests/chat-in-progress-settle.test.ts src/tests/stream-lock-leak.test.ts src/tests/stream-semaphore.test.ts src/tests/waf-isolation.test.ts src/tests/legacy-completions.test.ts src/tests/response-format.test.ts src/tests/cors-headers.test.ts src/tests/stream-lock-terminal-cancel.test.ts src/tests/parser-truncated-tool-call.test.ts src/tests/metrics-prometheus.test.ts src/tests/context-death-cleanup.test.ts src/tests/personalization-direct-fetch.test.ts src/tests/completion-headers-parity.test.ts src/tests/parallel-escape-repro.test.ts src/tests/chat-in-progress-scale-cost.test.ts src/tests/temp-chat-mode.test.ts src/tests/profile-reset.test.ts src/tests/error-classification.test.ts src/tests/quota-cooldown.test.ts src/tests/account-ready-gate.test.ts src/tests/tool-cap-notice.test.ts src/tests/interaction-flows.test.ts src/tests/headers-gate-cooldown-fallback.test.ts src/tests/unwrapped-tool-calls.test.ts src/tests/anthropic.test.ts src/tests/sync-clients.test.ts src/tests/sync-top10-clients.test.ts src/tests/cli.test.ts src/tests/paths.test.ts src/tests/tool-manifest.test.ts src/tests/tui.test.ts src/tests/tui-settings.test.ts src/tests/personalization-deadline.test.ts src/tests/update-cli.test.ts src/tests/batch-accounts.test.ts",
     "test:live": "tsx --test src/tests/concurrentChat.test.ts src/tests/agenticStress.test.ts",
     "benchmark:proxy": "npx tsx src/benchmarks/proxy-baseline.ts",
     "typecheck": "tsc --noEmit"

package/src/api/server.ts

@@ -3,7 +3,8 @@ import net from "node:net";
 import { v4 as uuidv4 } from "uuid";
 import { Hono, type Context } from "hono";
 import { serve } from "@hono/node-server";
-import { config } from "../core/config.js";
+import { config, getRuntimeChatMode, setRuntimeChatMode } from "../core/config.js";
+import { saveTuiSettings } from "../tui/settings.ts";
 import { metrics } from "../core/metrics.js";
 import { logger, maskEmail } from "../core/logger.js";
 import { MemoryCache } from "../cache/memory-cache.js";
@@ -248,7 +249,22 @@ app.post("/v1/upload", uploadFile);
 app.post("/v1/images/generations", imagesGenerations);
 app.post("/v1/videos/generations", videosGenerations);
 app.get("/v1/tasks/status/:taskId", videoTaskStatus);
-
+app.get("/v1/chat/mode", (c) => {
+  const error = verifyApiKey(c);
+  if (error) return error;
+  return c.json({ mode: getRuntimeChatMode() });
+});
+app.post("/v1/chat/mode", async (c) => {
+  const error = verifyApiKey(c);
+  if (error) return error;
+  const body = (await c.req.json().catch(() => ({}))) as { mode?: string };
+  if (!body.mode) {
+    return c.json({ error: { message: "Field 'mode' is required" } }, 400);
+  }
+  const updated = setRuntimeChatMode(body.mode);
+  saveTuiSettings({ chat: { mode: updated } });
+  return c.json({ success: true, mode: updated });
+});
 // OpenAI Responses API compatible routes
 app.route("", responsesApp);
 app.route("", anthropicApp);

package/src/core/accounts.ts

@@ -122,7 +122,137 @@ export function invalidateAccountsCache(): void {
   accountsCache = null;
   accountsCacheTime = 0;
 }
+export interface BatchAccountEntry {
+  email: string;
+  password: string;
+}
+
+export function parseBatchAccounts(rawInput: string): {
+  entries: BatchAccountEntry[];
+  invalid: string[];
+} {
+  if (!rawInput || typeof rawInput !== "string") {
+    return { entries: [], invalid: [] };
+  }
+
+  let text = rawInput.trim();
+  // Strip optional QWEN_ACCOUNTS= prefix and surrounding quotes
+  text = text.replace(/^QWEN_ACCOUNTS\s*=\s*["']?/i, "").replace(/["']?\s*$/i, "");
+
+  const lines = text.split(/\r?\n/);
+  const entries: BatchAccountEntry[] = [];
+  const invalid: string[] = [];
+  const seenEmails = new Set<string>();
+
+  for (const rawLine of lines) {
+    const line = rawLine.trim();
+    if (!line || line.startsWith("#") || line.startsWith("//")) continue;
+
+    // Check if line contains multiple accounts separated by comma or semicolon
+    const segments =
+      (line.includes(",") || line.includes(";")) && (line.match(/@/g) || []).length > 1
+        ? line.split(/[,;]/).map((s) => s.trim()).filter(Boolean)
+        : [line];
+
+    for (const seg of segments) {
+      let email = "";
+      let password = "";
+
+      if (seg.includes("---")) {
+        const parts = seg.split("---");
+        email = parts[0].trim();
+        password = parts.slice(1).join("---").trim();
+      } else if (seg.includes("\t")) {
+        const parts = seg.split("\t");
+        email = parts[0].trim();
+        password = parts.slice(1).join("\t").trim();
+      } else if (seg.includes(" | ")) {
+        const parts = seg.split(" | ");
+        email = parts[0].trim();
+        password = parts.slice(1).join(" | ").trim();
+      } else if (seg.includes(":")) {
+        const colonIdx = seg.indexOf(":");
+        email = seg.slice(0, colonIdx).trim();
+        password = seg.slice(colonIdx + 1).trim();
+      } else if (seg.includes(",")) {
+        const commaIdx = seg.indexOf(",");
+        email = seg.slice(0, commaIdx).trim();
+        password = seg.slice(commaIdx + 1).trim();
+      } else {
+        invalid.push(seg);
+        continue;
+      }
+
+      // Validate email format
+      const emailRegex = /^[^\s@]+@[^\s@]+\.[^\s@]+$/;
+      if (!email || !password || !emailRegex.test(email)) {
+        invalid.push(seg);
+        continue;
+      }
+
+      const normalizedEmail = email.toLowerCase();
+      if (!seenEmails.has(normalizedEmail)) {
+        seenEmails.add(normalizedEmail);
+        entries.push({ email, password });
+      }
+    }
+  }
+
+  return { entries, invalid };
+}
 
+export function addAccountsBatch(
+  entries: BatchAccountEntry[],
+): { added: QwenAccount[]; skipped: string[]; invalid: string[] } {
+  const db = getDatabase();
+  const added: QwenAccount[] = [];
+  const skipped: string[] = [];
+  const invalid: string[] = [];
+
+  const existingEmails = new Set(
+    (db.prepare("SELECT email FROM accounts").all() as Array<{ email: string }>).map((r) =>
+      r.email.toLowerCase(),
+    ),
+  );
+
+  const insertStmt = db.prepare(
+    "INSERT INTO accounts (id, email, password) VALUES (?, ?, ?)",
+  );
+
+  const insertBatch = db.transaction((items: BatchAccountEntry[]) => {
+    for (const item of items) {
+      const email = item.email.trim();
+      const password = item.password;
+      if (!email || !password) {
+        invalid.push(email || "(empty)");
+        continue;
+      }
+
+      if (existingEmails.has(email.toLowerCase())) {
+        skipped.push(email);
+        continue;
+      }
+
+      const newAccount: QwenAccount = {
+        id: crypto.randomUUID(),
+        email,
+        password,
+      };
+
+      insertStmt.run(newAccount.id, newAccount.email, encrypt(newAccount.password));
+      existingEmails.add(email.toLowerCase());
+      added.push(newAccount);
+    }
+  });
+
+  insertBatch(entries);
+
+  if (added.length > 0) {
+    invalidateAccountsCache();
+  }
+
+  return { added, skipped, invalid };
+}
 export function addAccount(
   email: string,
   password: string,

package/src/core/config.ts

@@ -34,7 +34,18 @@ const envSchema = z
     // temp chat (chat_mode:"local") for every request and sends the full
     // history inline (OpenAI standard). Temp chats are ephemeral and never
     // appear in the account's chat list (live-probed).
-    QWEN_CHAT_MODE: z.enum(["thread", "temp", "temp-thread"]).default("thread"),
+    QWEN_CHAT_MODE: z
+      .enum([
+        "thread",
+        "temp",
+        "temp-thread",
+        "thread-temp",
+        "stateless",
+        "stateles",
+        "stateless-temp",
+        "stateles-temp",
+      ])
+      .default("thread"),
     PLAYWRIGHT_HEADLESS: z.string().default("true"),
     PLAYWRIGHT_BROWSER: z
       .enum(["chromium", "chrome", "edge"])
@@ -362,7 +373,7 @@ export const config = {
     /** When true, all requests (personalization, models, media, chat) route exclusively through the browser page (no direct Node fetch). */
     browserOnlyFetch: env.QWEN_BROWSER_ONLY_FETCH !== "false",
     mapOpenAiModels: env.QWEN_MAP_OPENAI_MODELS !== "false",
-    chatMode: env.QWEN_CHAT_MODE,
+    chatMode: normalizeChatMode(env.QWEN_CHAT_MODE),
     maxPromptBytes: Math.max(0, parseInt(env.QWEN_MAX_PROMPT_BYTES)),
     maxPersonalizationBytes: Math.max(
       0,
@@ -388,5 +399,85 @@ export const config = {
 
 export type Config = typeof config;
 
-/** Conversation mode: thread-native reuse vs ephemeral temp chat per request vs temp-thread (ephemeral continuous session). */
-export type ChatMode = "thread" | "temp" | "temp-thread";
+/**
+ * Conversation modes:
+ * - "thread": thread-native delta (~1KB), saved in Qwen account sidebar (default)
+ * - "thread-temp" (alias "temp-thread"): thread-native delta (~1KB), ephemeral local chat (not saved)
+ * - "stateless": full history re-sent every turn (OpenAI standard), saved in Qwen account sidebar
+ * - "stateless-temp" (alias "temp"): full history re-sent every turn (OpenAI standard), ephemeral local chat (not saved)
+ */
+export type ChatMode =
+  | "thread"
+  | "thread-temp"
+  | "temp-thread"
+  | "stateless"
+  | "stateles"
+  | "stateless-temp"
+  | "stateles-temp"
+  | "temp";
+
+export function isStatelessChatMode(mode?: string | null): boolean {
+  if (!mode) return false;
+  const m = mode.trim().toLowerCase().replace(/_/g, "-");
+  return (
+    m === "stateless" ||
+    m === "stateles" ||
+    m === "stateless-temp" ||
+    m === "stateles-temp" ||
+    m === "temp"
+  );
+}
+
+export function isLocalChatMode(mode?: string | null): boolean {
+  if (!mode) return false;
+  const m = mode.trim().toLowerCase().replace(/_/g, "-");
+  return (
+    m === "stateless-temp" ||
+    m === "stateles-temp" ||
+    m === "temp" ||
+    m === "thread-temp" ||
+    m === "temp-thread"
+  );
+}
+
+export function normalizeChatMode(
+  mode?: string | null,
+): "thread" | "thread-temp" | "stateless" | "stateless-temp" {
+  if (!mode) return "thread";
+  const m = mode.trim().toLowerCase().replace(/_/g, "-");
+  if (m === "stateless-temp" || m === "stateles-temp" || m === "temp") {
+    return "stateless-temp";
+  }
+  if (m === "stateless" || m === "stateles") {
+    return "stateless";
+  }
+  if (m === "thread-temp" || m === "temp-thread") {
+    return "thread-temp";
+  }
+  return "thread";
+}
+
+export function setRuntimeChatMode(
+  mode?: string | null,
+): "thread" | "thread-temp" | "stateless" | "stateless-temp" {
+  const normalized = normalizeChatMode(mode);
+  (config.qwen as { chatMode: "thread" | "thread-temp" | "stateless" | "stateless-temp" }).chatMode = normalized;
+  return normalized;
+}
+
+export function getRuntimeChatMode(): "thread" | "thread-temp" | "stateless" | "stateless-temp" {
+  return config.qwen.chatMode as "thread" | "thread-temp" | "stateless" | "stateless-temp";
+}
+
+export function cycleNextChatMode(): "thread" | "thread-temp" | "stateless" | "stateless-temp" {
+  const current = getRuntimeChatMode();
+  const order: Array<"thread" | "thread-temp" | "stateless-temp" | "stateless"> = [
+    "thread",
+    "thread-temp",
+    "stateless-temp",
+    "stateless",
+  ];
+  const idx = order.indexOf(current);
+  const next = order[(idx + 1) % order.length];
+  return setRuntimeChatMode(next);
+}

package/src/index.ts

@@ -1,3 +1,6 @@
+try {
+  process.title = "QwenProxy";
+} catch {}
 process.env.DOTENV_CONFIG_QUIET = 'true'
 import dotenv from 'dotenv'
 import fs from 'node:fs'

package/src/routes/chat/context.ts

@@ -1,4 +1,4 @@
-import { config, type ChatMode } from "../../core/config.ts";
+import { config, type ChatMode, isStatelessChatMode } from "../../core/config.ts";
 import { ContextLengthExceededError, ValidationError } from "../../core/errors.ts";
 import { getModelContextWindow } from "../../core/model-registry.ts";
 import {
@@ -61,16 +61,16 @@ export async function buildFinalContext(
 
   const modelContextWindow = getModelContextWindow(modelId);
   const useThreadNative = true;
-  const isTempMode = chatMode === "temp";
+  const isStateless = isStatelessChatMode(chatMode);
   // A continuation is ANY evidence of a prior turn, not just a plain
   // role:"assistant" message. Tool-loop clients (Zed/Cline) can send history
   // with tool/function responses or assistant tool_calls but WITHOUT a plain
   // assistant entry; misclassifying those as a new session forced the FULL
   // history to be re-sent on every request (and every chat_in_progress retry)
   // instead of the thread-native delta.
-  // In temp mode EVERY request is a new (ephemeral) chat, so the whole history
-  // is always sent and no thread state is ever consulted.
-  const isNewSession = isTempMode
+  // In stateless modes (stateless, stateless-temp, temp) EVERY request is a
+  // new chat, so the whole history is always sent and no thread state is consulted.
+  const isNewSession = isStateless
     ? true
     : !messages.some(isContinuationMessage);
   const completeInstructions = [systemPrompt.trim(), toolInstructions.trim()]
@@ -83,7 +83,7 @@ export async function buildFinalContext(
   //    OR: this is a continuation (has assistant messages in history)
   // This prevents new IDE chats from accidentally reusing old Qwen chats
   // while still allowing continuations without explicit session_id
-  const allowThreadReuse = isTempMode
+  const allowThreadReuse = isStateless
     ? false
     : useThreadNative && (hasExplicitConversationKey || !isNewSession); // has assistant messages = continuation of existing chat
 
@@ -91,7 +91,7 @@ export async function buildFinalContext(
   // an explicit conversation key. Otherwise, generate an ephemeral ID for
   // logging/metrics only (not used for thread reuse). Temp mode never persists
   // a thread, so it has no session id.
-  const sessionId = isTempMode
+  const sessionId = isStateless
     ? null
     : (conversationKey || useThreadNative)
       ? deriveSessionId(
@@ -110,7 +110,7 @@ export async function buildFinalContext(
   // Thread-native: send full history when Qwen has no context yet, but preserve
   // tool-result deltas because the upstream parent chain already owns the call.
   // Temp mode: always send the FULL history (OpenAI standard).
-  const baseActivePrompt = isTempMode
+  const baseActivePrompt = isStateless
     ? prompt
     : (!existingThread && !hasTrailingToolResult ? prompt : currentPrompt) ||
       prompt;
@@ -179,7 +179,7 @@ export async function buildFinalContext(
     isNewSession,
     useThreadNative,
     // Thread state is only persisted in thread mode (temp chats are ephemeral).
-    updateLogicalThread: isTempMode ? false : useThreadNative,
+    updateLogicalThread: isStateless ? false : useThreadNative,
     chatMode,
     isThinkingModel,
     estimatedTokens,

package/src/routes/chat/index.ts

@@ -20,7 +20,7 @@ import {
   handleChatCompletionsError,
   type AssistantCompleteEvent,
 } from "./streaming.ts";
-import { config, type ChatMode } from "../../core/config.ts";
+import { config, type ChatMode, normalizeChatMode } from "../../core/config.ts";
 import { logger } from "../../core/logger.ts";
 import { metrics } from "../../core/metrics.ts";
 import { getContextMeterHeaders, type ContextMeterMode } from "../../services/context-meter.ts";
@@ -50,11 +50,20 @@ function formatTimingHeader(timings: Record<string, number>): string {
  * else silently uses the configured default.
  */
 function resolveChatMode(headerValue: string | undefined): ChatMode {
-  if (headerValue === "thread" || headerValue === "temp" || headerValue === "temp-thread") {
-    return headerValue;
-  }
-  if (headerValue === "temp_thread") {
-    return "temp-thread";
+  if (headerValue) {
+    const m = headerValue.trim().toLowerCase().replace(/_/g, "-");
+    if (
+      m === "thread" ||
+      m === "thread-temp" ||
+      m === "temp-thread" ||
+      m === "stateless" ||
+      m === "stateles" ||
+      m === "stateless-temp" ||
+      m === "stateles-temp" ||
+      m === "temp"
+    ) {
+      return normalizeChatMode(m);
+    }
   }
   return config.qwen.chatMode;
 }

package/src/services/qwen-chat-pool.ts

@@ -1,6 +1,6 @@
 import crypto from "crypto";
 import { getQwenHeaders, isAuthMockEnabled } from "./auth-playwright.ts";
-import { config, type ChatMode } from "../core/config.ts";
+import { config, type ChatMode, isLocalChatMode, isStatelessChatMode } from "../core/config.ts";
 import { logger, isToolcallDebugEnabled } from "../core/logger.ts";
 import {
   computeQuotaCooldownMs,
@@ -83,9 +83,8 @@ export function buildChatNewBody(
     project_id: "",
     timestamp: Date.now(),
     chat_type: "t2t",
-    // thread → normal (persisted), temp / temp-thread → local (ephemeral, not listed).
-    chat_mode:
-      chatMode === "temp" || chatMode === "temp-thread" ? "local" : "normal",
+    // normal: thread & stateless (persisted in Qwen); local: thread-temp & stateless-temp (ephemeral)
+    chat_mode: isLocalChatMode(chatMode) ? "local" : "normal",
   };
 }
 
@@ -201,7 +200,7 @@ export async function acquireNewQwenChatSession(
   accountId?: string,
   chatMode: ChatMode = "thread",
 ): Promise<{ chatId: string; leasedFromPool: boolean }> {
-  if (isQwenChatPoolEnabled() && chatMode !== "temp") {
+  if (isQwenChatPoolEnabled() && !isStatelessChatMode(chatMode)) {
     const key = chatPoolKey(accountId, model);
     const pooled = precreatedChatSessions.get(key);
     const chatId = pooled?.shift();

package/src/services/qwen.ts

@@ -12,7 +12,7 @@ import {
 } from "../core/errors.ts";
 import { buildQwenRequestHeaders } from "./qwen-headers.ts";
 import { qwenOrigin, qwenUrl } from "./qwen-url.ts";
-import { config, type ChatMode } from "../core/config.ts";
+import { config, type ChatMode, isLocalChatMode } from "../core/config.ts";
 import { logger } from "../core/logger.ts";
 import { estimateTokenCount } from "../utils/context-truncation.ts";
 import type {
@@ -2757,10 +2757,7 @@ async function createQwenStreamInternal(
     chatId: chatSessionId || null,
     parentId: actualParentId ?? "",
     chat_id: chatSessionId || null,
-    chat_mode:
-      options?.chatMode === "temp" || options?.chatMode === "temp-thread"
-        ? "local"
-        : "normal",
+    chat_mode: isLocalChatMode(options?.chatMode) ? "local" : "normal",
     model: model,
     parent_id: actualParentId,
     messages: [

package/src/tui/app.ts

@@ -26,6 +26,7 @@ import { SyncView } from "./views/sync-view.ts";
 import { StorageView } from "./views/storage-view.ts";
 import { AccountsView } from "./views/accounts-view.ts";
 import { LogsView } from "./views/logs-view.ts";
+import { setRuntimeChatMode } from "../core/config.ts";
 import { loadTuiSettings, saveTuiSettings } from "./settings.ts";
 export class TuiApp {
   private screen: Screen;
@@ -39,6 +40,10 @@ export class TuiApp {
   private renderScheduled = false;
   constructor(initialTab?: number) {
     this.screen = new Screen();
+    const saved = loadTuiSettings();
+    if (saved.chat?.mode) {
+      setRuntimeChatMode(saved.chat.mode);
+    }
 
     this.views = [
       new StatusView(),
@@ -49,16 +54,10 @@ export class TuiApp {
       new LogsView(),
     ];
 
-    let resolvedTab = initialTab;
-    if (!resolvedTab || isNaN(resolvedTab)) {
-      const saved = loadTuiSettings();
-      if (saved.lastTab && saved.lastTab >= 1 && saved.lastTab <= 6) {
-        resolvedTab = saved.lastTab;
-      } else {
-        resolvedTab = 1;
-      }
+    let resolvedTab = initialTab ?? (saved.lastTab && saved.lastTab >= 1 && saved.lastTab <= 6 ? saved.lastTab : 1);
+    if (isNaN(resolvedTab)) {
+      resolvedTab = 1;
     }
-
     const tabIdx = Math.max(0, Math.min(this.views.length - 1, resolvedTab - 1));
     this.activeViewIndex = tabIdx;
   }

package/src/tui/index.ts

@@ -23,6 +23,9 @@ function parseInitialTab(): number | undefined {
 
   return undefined;
 }
+try {
+  process.title = "QwenProxy";
+} catch {}
 
 async function main() {
   if (!process.stdout.isTTY || !process.stdin.isTTY) {

package/src/tui/proxy-client.ts

@@ -190,10 +190,11 @@ export async function fetchProxyStatus(): Promise<ProxyStatusSnapshot> {
   const cacheBytesSaved = lastMetricsData?.cache?.bytesSaved;
 
   return {
-    online,
+    online: lastOnlineState,
     port,
     host,
-    overallStatus,
+    chatMode: config.qwen.chatMode as any,
+    overallStatus: lastOverallStatus,
     uptimeSeconds,
     rssMb,
     systemMemoryPct,

package/src/tui/screen.ts

@@ -47,13 +47,20 @@ export class Screen {
     if (!process.stdout.isTTY || !process.stdin.isTTY) {
       return false;
     }
-
+    try {
+      process.title = "QwenProxy";
+    } catch {}
     this.active = true;
     this.prevRenderedRows = [];
-    // Switch to alternate screen buffer, clear screen, hide cursor, and enable mouse tracking
+    // Switch to alternate screen buffer, clear screen, hide cursor, enable mouse tracking, and set terminal title
     ServerManager.getInstance().withTuiRendering(() => {
       process.stdout.write(
-        ANSI.enterAltScreen + "\x1b[2J" + ANSI.cursorHome + ANSI.hideCursor + ANSI.enableMouse,
+        ANSI.enterAltScreen +
+          "\x1b[2J" +
+          ANSI.cursorHome +
+          ANSI.hideCursor +
+          ANSI.enableMouse +
+          ANSI.setTitle("QwenProxy"),
       );
     });
     // Setup raw keyboard input

package/src/tui/settings.ts

@@ -12,6 +12,7 @@ export interface TuiSettings {
   chat?: {
     model?: string;
     effort?: "high" | "medium" | "low";
+    mode?: "thread" | "thread-temp" | "stateless" | "stateless-temp";
   };
   logs?: {
     filter?: "all" | "warn" | "error";
@@ -23,6 +24,7 @@ const defaultSettings: TuiSettings = {
   chat: {
     model: "qwen3.8-max",
     effort: "high",
+    mode: "thread",
   },
   logs: {
     filter: "all",

package/src/tui/theme.ts

@@ -18,6 +18,7 @@ export const ANSI = {
   exitAltScreen: "\x1b[?1049l",
   enableMouse: "\x1b[?1000h\x1b[?1002h\x1b[?1003h\x1b[?1006h",
   disableMouse: "\x1b[?1006l\x1b[?1005l\x1b[?1004l\x1b[?1003l\x1b[?1002l\x1b[?1000l\x1b[?1015l",
+  setTitle: (title: string) => `\x1b]0;${title}\x07`,
 };
 
 export const theme = {

package/src/tui/types.ts

@@ -15,6 +15,7 @@ export interface ProxyStatusSnapshot {
   online: boolean;
   port: number;
   host: string;
+  chatMode?: "thread" | "thread-temp" | "stateless" | "stateless-temp";
   overallStatus?: string;
   uptimeSeconds?: number;
   rssMb?: number;