fix(providers): refresh dated Token Plan catalogue and metadata

docs-site/src/content/docs/fr/guides/providers.md

@@ -364,7 +364,7 @@ promotionnels de Cline ne sont accessibles que dans l'IDE ou la CLI Cline, pas p
 | Z.AI (GLM Coding) | `https://api.z.ai/api/coding/paas/v4` |
 | Zhipu AI (BigModel) | `https://open.bigmodel.cn/api/paas/v4` |
 | [BigModel Coding Plan — Responses (liste statique)](/guides/providers/#bigmodel-coding-plan-over-responses) | `https://open.bigmodel.cn/api/v1` |
-| Qwen Cloud | Forfait à jetons (par défaut) : `https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` · Facturation à l'usage : `https://dashscope.aliyuncs.com/compatible-mode/v1` · ou personnalisé |
+| Qwen Cloud | Forfait à jetons (par défaut) : `https://token-plan.maas.qwencloudapi.com/compatible-mode/v1` · Facturation à l'usage : `https://dashscope.aliyuncs.com/compatible-mode/v1` · ou personnalisé |
 | Tencent Cloud Coding Plan | `https://api.lkeap.cloud.tencent.com/coding/v3` |
 | SiliconFlow | `https://api.siliconflow.cn/v1` |
 | Volcengine Ark · Coding Plan · Agent Plan | `https://ark.cn-beijing.volces.com/api/v3` · `https://ark.cn-beijing.volces.com/api/coding/v3` · `https://ark.cn-beijing.volces.com/api/plan/v3` |

docs-site/src/content/docs/guides/providers.md

@@ -739,7 +739,7 @@ region-pinned EU routes, is at [opper.ai/models](https://opper.ai/models). Opper
 | Z.AI (GLM Coding) | `https://api.z.ai` — Responses at `/api/v1/responses` by default; Chat Completions at `/api/coding/paas/v4/chat/completions` per model through `modelAdapters` |
 | Zhipu AI (BigModel) | `https://open.bigmodel.cn/api/paas/v4` |
 | BigModel Coding Plan (Responses, static roster) | `https://open.bigmodel.cn/api/v1` |
-| Qwen Cloud | Token plan (default): `https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` · Pay as you go: `https://dashscope.aliyuncs.com/compatible-mode/v1` · or Custom |
+| Qwen Cloud | Token plan (default): `https://token-plan.maas.qwencloudapi.com/compatible-mode/v1` · Pay as you go: `https://dashscope.aliyuncs.com/compatible-mode/v1` · or Custom |
 | Tencent Cloud Coding Plan | `https://api.lkeap.cloud.tencent.com/coding/v3` |
 | SiliconFlow | `https://api.siliconflow.cn/v1` |
 | Volcengine Ark · Coding Plan · Agent Plan | `https://ark.cn-beijing.volces.com/api/v3` · `https://ark.cn-beijing.volces.com/api/coding/v3` · `https://ark.cn-beijing.volces.com/api/plan/v3` |
@@ -754,6 +754,27 @@ region-pinned EU routes, is at [opper.ai/models](https://opper.ai/models). Opper
 | Cloudflare AI Gateway | `https://gateway.ai.cloudflare.com/v1/{account-id}/{gateway}/anthropic` |
 | …and more | opencode zen, Vercel AI Gateway, Venice, NanoGPT, Synthetic, Qianfan, Alibaba, Parallel, ZenMux, LiteLLM |
 
+**Alibaba Token Plan** provides Beijing Personal Edition and International Team Edition
+presets. Both regions offer both subscription tiers. The presets keep static chat-model
+lists because upstream discovery also includes non-chat models. Existing selections and
+explicit International endpoint overrides are preserved.
+
+Both presets include `deepseek-v4.1-flash-1017` with text/image input, text output,
+1,000,000 context and `low`/`high`/`max` reasoning (default `high`). Its 393,216 output
+budget follows the existing published-budget normalization, not a measured boundary;
+the lab lists 384,000. Strict schemas reject and `json_object` is unverified for this
+exact ID, so Chat requests omit `response_format`. This is best-effort text, not
+guaranteed JSON. The bare model's verified JSON downgrade and existing Responses
+pins are unchanged. Beijing also includes now-listed `deepseek-v4-pro-0813`;
+International omits retired `deepseek-v3.2` and bare `deepseek-v4-flash` but retains
+`deepseek-v4-flash-0731`. Both presets keep the hosts their official quick starts
+document: Beijing `https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` and
+International `https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`.
+The Qwen Cloud preset uses the separate Token Plan host that Qwen Cloud documents above.
+See the [1017 model card](https://www.qianwenai.com/models/deepseek-v4.1-flash-1017),
+[Chat reference](https://docs.qwencloud.com/api-reference/chat/openai-chat) and
+[Personal overview](https://help.aliyun.com/en/model-studio/token-plan-personal-overview).
+
 **OpenGateway** is an OpenAI-compatible gateway operated by Sionic AI at
 `https://apis.opengateway.ai/v1`. Its public catalog contains about 80 active models
 (as verified on 2026-10-02); the preset automatically refreshes the live model list from

docs-site/src/content/docs/ja/guides/providers.md

@@ -260,7 +260,7 @@ Cline IDE/CLI のみで API からは使えません。`minimax/minimax-m2.5` 
 | Z.AI (GLM Coding) | `https://api.z.ai/api/coding/paas/v4` |
 | Zhipu AI (BigModel) | `https://open.bigmodel.cn/api/paas/v4` |
 | [BigModel Coding Plan — Responses (静的モデル一覧)](/guides/providers/#bigmodel-coding-plan-over-responses) | `https://open.bigmodel.cn/api/v1` |
-| Qwen Cloud | トークンプラン(デフォルト): `https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` · 従量課金: `https://dashscope.aliyuncs.com/compatible-mode/v1` · またはカスタム |
+| Qwen Cloud | トークンプラン(デフォルト): `https://token-plan.maas.qwencloudapi.com/compatible-mode/v1` · 従量課金: `https://dashscope.aliyuncs.com/compatible-mode/v1` · またはカスタム |
 | Tencent Cloud Coding Plan | `https://api.lkeap.cloud.tencent.com/coding/v3` |
 | SiliconFlow | `https://api.siliconflow.cn/v1` |
 | Volcengine Ark · Coding Plan · Agent Plan | `https://ark.cn-beijing.volces.com/api/v3` · `https://ark.cn-beijing.volces.com/api/coding/v3` · `https://ark.cn-beijing.volces.com/api/plan/v3` |

docs-site/src/content/docs/ko/guides/providers.md

@@ -265,7 +265,7 @@ Cline IDE/CLI에서만 제공되며 API로는 사용할 수 없습니다. `minim
 | Z.AI (GLM Coding) | `https://api.z.ai/api/coding/paas/v4` |
 | Zhipu AI (BigModel) | `https://open.bigmodel.cn/api/paas/v4` |
 | [BigModel Coding Plan — Responses (정적 모델 목록)](/guides/providers/#bigmodel-coding-plan-over-responses) | `https://open.bigmodel.cn/api/v1` |
-| Qwen Cloud | Token plan(기본): `https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` · 종량제: `https://dashscope.aliyuncs.com/compatible-mode/v1` · 또는 사용자 지정 |
+| Qwen Cloud | Token plan(기본): `https://token-plan.maas.qwencloudapi.com/compatible-mode/v1` · 종량제: `https://dashscope.aliyuncs.com/compatible-mode/v1` · 또는 사용자 지정 |
 | Tencent Cloud Coding Plan | `https://api.lkeap.cloud.tencent.com/coding/v3` |
 | SiliconFlow | `https://api.siliconflow.cn/v1` |
 | Volcengine Ark · Coding Plan · Agent Plan | `https://ark.cn-beijing.volces.com/api/v3` · `https://ark.cn-beijing.volces.com/api/coding/v3` · `https://ark.cn-beijing.volces.com/api/plan/v3` |

docs-site/src/content/docs/ru/guides/providers.md

@@ -275,7 +275,7 @@ opencodex поставляется с 104 встроенными пресета
 | Z.AI (GLM Coding) | `https://api.z.ai/api/coding/paas/v4` |
 | Zhipu AI (BigModel) | `https://open.bigmodel.cn/api/paas/v4` |
 | [BigModel Coding Plan — Responses (статический список)](/guides/providers/#bigmodel-coding-plan-over-responses) | `https://open.bigmodel.cn/api/v1` |
-| Qwen Cloud | Token plan (по умолчанию): `https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` · Pay as you go: `https://dashscope.aliyuncs.com/compatible-mode/v1` · или Custom |
+| Qwen Cloud | Token plan (по умолчанию): `https://token-plan.maas.qwencloudapi.com/compatible-mode/v1` · Pay as you go: `https://dashscope.aliyuncs.com/compatible-mode/v1` · или Custom |
 | Tencent Cloud Coding Plan | `https://api.lkeap.cloud.tencent.com/coding/v3` |
 | SiliconFlow | `https://api.siliconflow.cn/v1` |
 | Volcengine Ark · Coding Plan · Agent Plan | `https://ark.cn-beijing.volces.com/api/v3` · `https://ark.cn-beijing.volces.com/api/coding/v3` · `https://ark.cn-beijing.volces.com/api/plan/v3` |

docs-site/src/content/docs/tr/guides/providers.md

@@ -402,7 +402,7 @@ yalnızca Cline IDE/CLI içinde mevcuttur; `minimax/minimax-m2.5` belgelenmiş A
 | Z.AI (GLM Kodlama) | `https://api.z.ai/api/coding/paas/v4` |
 | Zhipu AI (BigModel) | `https://open.bigmodel.cn/api/paas/v4` |
 | [BigModel Coding Plan — Responses (statik model listesi)](/guides/providers/#bigmodel-coding-plan-over-responses) | `https://open.bigmodel.cn/api/v1` |
-| Qwen Cloud | Token planı (varsayılan): `https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` · Kullandıkça öde: `https://dashscope.aliyuncs.com/compatible-mode/v1` · veya Özel |
+| Qwen Cloud | Token planı (varsayılan): `https://token-plan.maas.qwencloudapi.com/compatible-mode/v1` · Kullandıkça öde: `https://dashscope.aliyuncs.com/compatible-mode/v1` · veya Özel |
 | Tencent Cloud Coding Plan | `https://api.lkeap.cloud.tencent.com/coding/v3` |
 | SiliconFlow | `https://api.siliconflow.cn/v1` |
 | Volcengine Ark · Coding Plan · Agent Plan | `https://ark.cn-beijing.volces.com/api/v3` · `https://ark.cn-beijing.volces.com/api/coding/v3` · `https://ark.cn-beijing.volces.com/api/plan/v3` |

docs-site/src/content/docs/zh-cn/guides/providers.md

@@ -254,7 +254,7 @@ Cline IDE/CLI 中提供，不能通过 API 使用；`minimax/minimax-m2.5` 是
 | Z.AI (GLM Coding) | `https://api.z.ai/api/coding/paas/v4` |
 | 智谱 AI (BigModel) | `https://open.bigmodel.cn/api/paas/v4` |
 | [BigModel Coding Plan — Responses (静态模型列表)](/guides/providers/#bigmodel-coding-plan-over-responses) | `https://open.bigmodel.cn/api/v1` |
-| Qwen Cloud | Token plan（默认）: `https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` · 按量付费: `https://dashscope.aliyuncs.com/compatible-mode/v1` · 或自定义 |
+| Qwen Cloud | Token plan（默认）: `https://token-plan.maas.qwencloudapi.com/compatible-mode/v1` · 按量付费: `https://dashscope.aliyuncs.com/compatible-mode/v1` · 或自定义 |
 | 腾讯云 Coding Plan | `https://api.lkeap.cloud.tencent.com/coding/v3` |
 | SiliconFlow | `https://api.siliconflow.cn/v1` |
 | 火山方舟 · Coding Plan · Agent Plan | `https://ark.cn-beijing.volces.com/api/v3` · `https://ark.cn-beijing.volces.com/api/coding/v3` · `https://ark.cn-beijing.volces.com/api/plan/v3` |

docs-site/src/content/docs/zh-tw/guides/providers.md

@@ -316,7 +316,7 @@ IDE／CLI，不透過 API；`minimax/minimax-m2.5` 是文件列出的 API 免費
 | Z.AI (GLM Coding) | `https://api.z.ai/api/coding/paas/v4` |
 | Zhipu AI (BigModel) | `https://open.bigmodel.cn/api/paas/v4` |
 | [BigModel Coding Plan — Responses (靜態模型清單)](/guides/providers/#bigmodel-coding-plan-over-responses) | `https://open.bigmodel.cn/api/v1` |
-| Qwen Cloud | Token plan（預設）：`https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` · pay as you go：`https://dashscope.aliyuncs.com/compatible-mode/v1` · 或 Custom |
+| Qwen Cloud | Token plan（預設）：`https://token-plan.maas.qwencloudapi.com/compatible-mode/v1` · pay as you go：`https://dashscope.aliyuncs.com/compatible-mode/v1` · 或 Custom |
 | Tencent Cloud Coding Plan | `https://api.lkeap.cloud.tencent.com/coding/v3` |
 | SiliconFlow | `https://api.siliconflow.cn/v1` |
 | Volcengine Ark · Coding Plan · Agent Plan | `https://ark.cn-beijing.volces.com/api/v3` · `https://ark.cn-beijing.volces.com/api/coding/v3` · `https://ark.cn-beijing.volces.com/api/plan/v3` |

src/providers/base-url-choices.ts

@@ -1,9 +1,11 @@
 /**
  * Qwen Cloud OpenAI-compatible endpoint presets.
  * Token plan is the GUI default; pay-as-you-go and custom are selectable.
+ * Token plan host per https://docs.qwencloud.com/token-plan/personal/token-plan-personal-quickstart
+ * (checked 2026-10-23); a saved older host stays usable as a custom URL.
  */
 export const QWEN_CLOUD_TOKEN_PLAN_BASE_URL =
-  "https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1";
+  "https://token-plan.maas.qwencloudapi.com/compatible-mode/v1";
 export const QWEN_CLOUD_PAYG_BASE_URL =
   "https://dashscope.aliyuncs.com/compatible-mode/v1";
 
@@ -23,10 +25,11 @@ export const QWEN_CLOUD_BASE_URL_CHOICES: readonly ProviderBaseUrlChoice[] = [
 
 /**
  * Alibaba Token Plan International (ap-southeast-1) endpoint presets.
- * Same product as the Beijing Token Plan but for international accounts.
- * Note: ALIBABA_INTL_TOKEN_PLAN_BASE_URL intentionally duplicates
- * QWEN_CLOUD_TOKEN_PLAN_BASE_URL — same host, different product branding
- * and model lineup. Kept as a separate constant for clarity.
+ * Separate regional account and billing destination; both regions offer both tiers.
+ * Alibaba Cloud's own quick start still documents this host (checked 2026-10-23):
+ * https://www.alibabacloud.com/help/en/model-studio/token-plan-personal-quick-start
+ * Qwen Cloud documents a different Token Plan host, so the two constants are
+ * independent and must not be assumed equal.
  */
 export const ALIBABA_INTL_TOKEN_PLAN_BASE_URL =
   "https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1";

src/providers/derive.ts

@@ -3,6 +3,7 @@ import { COST_VENDOR_PRIORITY, getModelMetadata, getModelMetadataCaseInsensitive
 import { cloneFastWire } from "./fastwire";
 import { isAzureModelMetadataDestination, publishedAzureModelMetadata, type AzureModelMetadata } from "./azure-model-metadata";
 import { resolveModelPolicy } from "./resolved-model-policy";
+import { mapFill, nestedMapFill, stableUnion } from "./resolved-model-policy-merge";
 import {
   PROVIDER_REGISTRY,
   registryEntryForProviderDestination,
@@ -474,13 +475,22 @@ function applyVerbosityDefaults(prov: OcxProviderConfig, entry: ProviderRegistry
  * was skipped and the reasoning ladder was advertised without summary support — exactly the
  * inconsistency that makes Codex drop the inbound reasoning object.
  *
- * Deliberately narrow: reasoning-summary, effort, and replay-compatibility metadata only, via
+ * Deliberately narrow: reasoning-summary, effort, structured-output, and replay-compatibility metadata only, via
  * `registryEntryForProviderDestination`, which matches fixed key destinations and refuses
  * templated or overridable base URLs. A custom row keeps its own identity for everything else.
  */
 function enrichReasoningMetadataByDestination(prov: OcxProviderConfig): void {
   const destination = registryEntryForProviderDestination(prov);
   applyReasoningSummaryDefaults(prov, destination?.modelSupportsReasoningSummaries);
+  if (destination?.modelDefaultReasoningEfforts) {
+    prov.modelDefaultReasoningEfforts = mapFill(destination.modelDefaultReasoningEfforts, prov.modelDefaultReasoningEfforts)[0];
+  }
+  if (destination?.modelReasoningEffortMap) {
+    prov.modelReasoningEffortMap = nestedMapFill(destination.modelReasoningEffortMap, prov.modelReasoningEffortMap)[0];
+  }
+  if (destination?.noStructuredOutputModels?.length) {
+    prov.noStructuredOutputModels = stableUnion(destination.noStructuredOutputModels, prov.noStructuredOutputModels)[0];
+  }
   if (prov.dropResponsesReasoningItems === undefined && destination?.dropResponsesReasoningItems !== undefined) {
     prov.dropResponsesReasoningItems = destination.dropResponsesReasoningItems;
   }

src/providers/registry/entries-extended.ts

@@ -837,6 +837,8 @@ export const PROVIDER_REGISTRY_EXTENDED: readonly ProviderRegistryEntry[] = [
   {
     id: "alibaba-token-plan",
     label: "Alibaba Token Plan (Beijing)",
+    // Official Personal quick start (2026-10-23) still documents this host:
+    // https://help.aliyun.com/en/model-studio/token-plan-personal-quick-start
     baseUrl: "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1",
     adapter: "openai-chat",
     authKind: "key",
@@ -887,19 +889,23 @@ export const PROVIDER_REGISTRY_EXTENDED: readonly ProviderRegistryEntry[] = [
       "deepseek-v4-pro-0813": deepseekThinkingEffortsFor("deepseek-v4-pro-0813"),
       "deepseek-v4-flash-0731": deepseekThinkingEffortsFor("deepseek-v4-flash-0731"),
       "deepseek-v4.1-flash": deepseekThinkingEffortsFor("deepseek-v4.1-flash"),
+      "deepseek-v4.1-flash-1017": deepseekThinkingEffortsFor("deepseek-v4.1-flash-1017"),
     },
     modelReasoningEffortMap: {
       "deepseek-v4-pro": deepseekReasoningMapFor("deepseek-v4-pro"),
       "deepseek-v4-pro-0813": deepseekReasoningMapFor("deepseek-v4-pro-0813"),
       "deepseek-v4-flash-0731": deepseekReasoningMapFor("deepseek-v4-flash-0731"),
       "deepseek-v4.1-flash": deepseekReasoningMapFor("deepseek-v4.1-flash"),
+      "deepseek-v4.1-flash-1017": deepseekReasoningMapFor("deepseek-v4.1-flash-1017"),
     },
     // Probed 260915 on the plan gateway: json_object returns valid JSON, strict
     // json_schema is rejected 400 ("This response_format type is unavailable now")
     // in both thinking modes, so requests downgrade to json_object rather than
     // sending a schema the gateway refuses.
     noJsonSchemaModels: ["deepseek-v4.1-flash"],
-    modelDefaultReasoningEfforts: Object.fromEntries(QWEN38_FAMILY.map(id => [id, "xhigh"])),
+    // 261020 strict schema rejected; json_object unverified. Omit rather than invent a fallback.
+    noStructuredOutputModels: ["deepseek-v4.1-flash-1017"],
+    modelDefaultReasoningEfforts: { ...Object.fromEntries(QWEN38_FAMILY.map(id => [id, "xhigh"])), "deepseek-v4.1-flash-1017": "high" },
     directReasoningEffortModels: QWEN38_FAMILY,
     thinkingBudgetModels: ALIBABA_TOKEN_PLAN_QWEN_MODELS.filter(id => !QWEN38_FAMILY.includes(id)),
     preserveReasoningContentModels: ALIBABA_TOKEN_PLAN_PRESERVE_REASONING,
@@ -935,25 +941,27 @@ export const PROVIDER_REGISTRY_EXTENDED: readonly ProviderRegistryEntry[] = [
       "glm-5.3": ZAI_GLM_53_REASONING_EFFORTS,
       "deepseek-v4-pro": deepseekThinkingEffortsFor("deepseek-v4-pro"),
       "deepseek-v4-pro-0813": deepseekThinkingEffortsFor("deepseek-v4-pro-0813"),
-      "deepseek-v4-flash": deepseekThinkingEffortsFor("deepseek-v4-flash"),
       "deepseek-v4-flash-0731": deepseekThinkingEffortsFor("deepseek-v4-flash-0731"),
       "deepseek-v4.1-flash": deepseekThinkingEffortsFor("deepseek-v4.1-flash"),
+      "deepseek-v4.1-flash-1017": deepseekThinkingEffortsFor("deepseek-v4.1-flash-1017"),
     },
     modelReasoningEffortMap: {
       "deepseek-v4-pro": deepseekReasoningMapFor("deepseek-v4-pro"),
       "deepseek-v4-pro-0813": deepseekReasoningMapFor("deepseek-v4-pro-0813"),
-      "deepseek-v4-flash": deepseekReasoningMapFor("deepseek-v4-flash"),
       "deepseek-v4-flash-0731": deepseekReasoningMapFor("deepseek-v4-flash-0731"),
       "deepseek-v4.1-flash": deepseekReasoningMapFor("deepseek-v4.1-flash"),
+      "deepseek-v4.1-flash-1017": deepseekReasoningMapFor("deepseek-v4.1-flash-1017"),
     },
     // Same 260915 json_schema rejection probe as the Beijing entry.
     noJsonSchemaModels: ["deepseek-v4.1-flash"],
+    // 261020 strict schema rejected; json_object unverified. Omit rather than invent a fallback.
+    noStructuredOutputModels: ["deepseek-v4.1-flash-1017"],
     directReasoningEffortModels: QWEN38_FAMILY,
     thinkingBudgetModels: ALIBABA_INTL_TOKEN_PLAN_QWEN_MODELS.filter(id => !QWEN38_FAMILY.includes(id)),
     preserveReasoningContentModels: ALIBABA_TOKEN_PLAN_PRESERVE_REASONING,
     noVisionModels: ALIBABA_TOKEN_PLAN_NO_VISION,
-    noReasoningModels: ["kimi-k2.7-code", "kimi-k2.6", "kimi-k2.5", "deepseek-v3.2", "glm-5.1", "glm-5", "MiniMax-M2.5"],
-    modelDefaultReasoningEfforts: Object.fromEntries(QWEN38_FAMILY.map(id => [id, "xhigh"])),
+    noReasoningModels: ["kimi-k2.7-code", "kimi-k2.6", "kimi-k2.5", "glm-5.1", "glm-5", "MiniMax-M2.5"],
+    modelDefaultReasoningEfforts: { ...Object.fromEntries(QWEN38_FAMILY.map(id => [id, "xhigh"])), "deepseek-v4.1-flash-1017": "high" },
     promptCacheKey: true,
   },
   // NEEDS_HUMAN 2026-07-10: kept for config compatibility, but this is a dashboard URL,

src/providers/registry/model-ids.ts

@@ -115,6 +115,7 @@ export const REGISTRY_FIELD_MODEL_ID_ROLES = {
   noStopModels: NONE,
   noPenaltyModels: NONE,
   noJsonSchemaModels: NONE,
+  noStructuredOutputModels: NONE,
   parallelToolCalls: NONE,
   promptCacheKey: NONE,
   chatServiceTier: NONE,