diff --git a/services/discord-gateway/src/modules/ai-moderation/cultureLearner.ts b/services/discord-gateway/src/modules/ai-moderation/cultureLearner.ts
index e71d537..49b2c2a 100644
--- a/services/discord-gateway/src/modules/ai-moderation/cultureLearner.ts
+++ b/services/discord-gateway/src/modules/ai-moderation/cultureLearner.ts
@@ -87,6 +87,7 @@ hal dasar ke newbie tanpa judge."`;
       max_tokens: 500,
       temperature: 0.7, // Higher temp for summarization
       retries: 2,
+      stream: true, // router always streams SSE; non-stream waits for full body and times out
     });
 
     if (!completion) throw new Error("Empty response from LLM");
diff --git a/services/discord-gateway/src/modules/ai-moderation/llmCaller.ts b/services/discord-gateway/src/modules/ai-moderation/llmCaller.ts
index 4ed3eb2..f2f471a 100644
--- a/services/discord-gateway/src/modules/ai-moderation/llmCaller.ts
+++ b/services/discord-gateway/src/modules/ai-moderation/llmCaller.ts
@@ -79,6 +79,13 @@ export async function callModerationLLM(
             jsonResponse: { type: "json_object" },
             retries: 0,
             signal,
+            // Router (9router/omniroute) always streams SSE even when the
+            // request omits `stream`. In non-stream mode the OpenAI SDK waits
+            // for the FULL body before parsing, so slow/long upstream streams
+            // hit the 30s/60s timeout and abort mid-generation. Streaming mode
+            // consumes chunks incrementally — timeout only fires on a real
+            // stall. llmClient aggregates the stream into a ChatCompletion.
+            stream: true,
           });
 
           if (!completion)
diff --git a/services/discord-gateway/src/modules/ai-moderation/llmClient.ts b/services/discord-gateway/src/modules/ai-moderation/llmClient.ts
index 3969423..230b4a5 100644
--- a/services/discord-gateway/src/modules/ai-moderation/llmClient.ts
+++ b/services/discord-gateway/src/modules/ai-moderation/llmClient.ts
@@ -272,6 +272,7 @@ export async function llmVision(
     temperature: 0.1,
     top_p: 0.9,
     retries: 0,
+    stream: true, // router always streams SSE; non-stream waits for full body and times out
   });
 
   if (!completion) return null;
diff --git a/services/discord-gateway/src/modules/ai-moderation/userProfileLearner.ts b/services/discord-gateway/src/modules/ai-moderation/userProfileLearner.ts
index 18e4400..ac8d8aa 100644
--- a/services/discord-gateway/src/modules/ai-moderation/userProfileLearner.ts
+++ b/services/discord-gateway/src/modules/ai-moderation/userProfileLearner.ts
@@ -116,6 +116,7 @@ atau konten SARA, itu akan SANGAT tidak sesuai dengan karakternya dan patut dicu
       max_tokens: 2000,
       temperature: 0.7, // Higher temp for summarization
       retries: 2,
+      stream: true, // router always streams SSE; non-stream waits for full body and times out
     });
 
     if (!completion) throw new Error("Empty response from LLM");