diff --git a/src/modules/llms/server/listModels.dispatch.ts b/src/modules/llms/server/listModels.dispatch.ts
index e513c922f..056bb4227 100644
--- a/src/modules/llms/server/listModels.dispatch.ts
+++ b/src/modules/llms/server/listModels.dispatch.ts
@@ -29,7 +29,7 @@ import { openAIAccess } from './openai/openai.access';
 import { alibabaModelFilter, alibabaModelSort, alibabaModelToModelDescription } from './openai/models/alibaba.models';
 import { azureDeploymentFilter, azureDeploymentToModelDescription, azureParseFromDeploymentsAPI } from './openai/models/azure.models';
 import { chutesAIHeuristic, chutesAIModelsToModelDescriptions } from './openai/models/chutesai.models';
-import { deepseekModelFilter, deepseekModelSort, deepseekModelToModelDescription } from './openai/models/deepseek.models';
+import { deepseekInjectVariants, deepseekModelFilter, deepseekModelSort, deepseekModelToModelDescription } from './openai/models/deepseek.models';
 import { fastAPIHeuristic, fastAPIModels } from './openai/models/fastapi.models';
 import { fireworksAIHeuristic, fireworksAIModelsToModelDescriptions } from './openai/models/fireworksai.models';
 import { groqModelFilter, groqModelSortFn, groqModelToModelDescription } from './openai/models/groq.models';
@@ -345,9 +345,11 @@ function _listModelsCreateDispatch(access: AixAPI_Access, signal?: AbortSignal):
                 .sort(openAISortModels);
 
             case 'deepseek':
+              // [DeepSeek, 2025-12-01] Inject V3.2-Speciale variant via reduce
               return maybeModels
                 .filter(({ id }) => deepseekModelFilter(id))
                 .map(({ id }) => deepseekModelToModelDescription(id))
+                .reduce(deepseekInjectVariants, [] as ModelDescriptionSchema[])
                 .sort(deepseekModelSort);
 
             case 'groq':
diff --git a/src/modules/llms/server/openai/models/deepseek.models.ts b/src/modules/llms/server/openai/models/deepseek.models.ts
index e6a6c1ac7..3e2d3740c 100644
--- a/src/modules/llms/server/openai/models/deepseek.models.ts
+++ b/src/modules/llms/server/openai/models/deepseek.models.ts
@@ -8,10 +8,10 @@ import { fromManualMapping, ManualMappings } from '../../models.mappings';
 const _knownDeepseekChatModels: ManualMappings = [
   // [Models and Pricing](https://api-docs.deepseek.com/quick_start/pricing)
   // [List Models](https://api-docs.deepseek.com/api/list-models)
-  // [Release Notes - V3.2-Exp](https://api-docs.deepseek.com/news/news250929) - Released 2025-09-29
+  // [Release Notes - V3.2](https://api-docs.deepseek.com/news/news251201) - Released 2025-12-01
   {
     idPrefix: 'deepseek-reasoner',
-    label: 'DeepSeek V3.2-Exp (Reasoner)',
+    label: 'DeepSeek V3.2 (Reasoner)',
     description: 'Reasoning model with Chain-of-Thought capabilities, 128K context length. Supports JSON output and function calling.',
     contextWindow: 131072, // 128K
     interfaces: [LLM_IF_OAI_Chat, LLM_IF_OAI_Fn, LLM_IF_OAI_Json, LLM_IF_OAI_Reasoning],
@@ -21,7 +21,7 @@ const _knownDeepseekChatModels: ManualMappings = [
   },
   {
     idPrefix: 'deepseek-chat',
-    label: 'DeepSeek V3.2-Exp',
+    label: 'DeepSeek V3.2',
     description: 'General-purpose model with 128K context length. Supports JSON output and function calling.',
     contextWindow: 131072, // 128K
     interfaces: [LLM_IF_OAI_Chat, LLM_IF_OAI_Fn, LLM_IF_OAI_Json],
@@ -59,3 +59,28 @@ export function deepseekModelSort(a: ModelDescriptionSchema, b: ModelDescription
     return aIndex - bIndex;
   return a.id.localeCompare(b.id);
 }
+
+
+// [DeepSeek, 2025-12-01] V3.2-Speciale: Temporary endpoint until Dec 15, 2025 15:59 UTC
+// Thinking mode only (deepseek-reasoner), 128K max output, no JSON/tool calling
+export const DEEPSEEK_SPECIALE_HOST = 'https://api.deepseek.com/v3.2_speciale_expires_on_20251215';
+export const DEEPSEEK_SPECIALE_SUFFIX = '@speciale';
+
+const _hardcodedDeepseekVariants: { [modelId: string]: Partial<ModelDescriptionSchema> } = {
+  'deepseek-reasoner': {
+    id: 'deepseek-reasoner' + DEEPSEEK_SPECIALE_SUFFIX, // [DeepSeek, 2025-12-01] marker for dispatch routing (no idVariant - the @speciale suffix serves as both)
+    label: 'DeepSeek V3.2 Speciale',
+    description: 'V3.2-Speciale reasoning model. 128K max output, no JSON/tool calling. Expires Dec 15, 2025.',
+    interfaces: [LLM_IF_OAI_Chat, LLM_IF_OAI_Reasoning], // NO Fn, NO Json
+    // contextWindow: null,
+    // maxCompletionTokens: undefined, // default 64K, max 128K (higher than regular reasoner's 32K default)
+  },
+};
+
+export function deepseekInjectVariants(models: ModelDescriptionSchema[], model: ModelDescriptionSchema): ModelDescriptionSchema[] {
+  // [DeepSeek, 2025-12-01] Inject Speciale variant for deepseek-reasoner
+  if (_hardcodedDeepseekVariants[model.id])
+    models.push({ ...model, ..._hardcodedDeepseekVariants[model.id] });
+  models.push(model);
+  return models;
+}