feat(chat): add configurable streaming output

ai_knowledge_base_v2/apps/backend/app/api/admin_settings.py

@@ -212,6 +212,7 @@ def save_agent_runtime_config(
     model.presence_penalty = payload.presencePenalty
     model.frequency_penalty = payload.frequencyPenalty
     model.max_token = payload.maxToken
+    model.stream_enabled = payload.streamEnabled
     db.add(model)
     OperationLogService.write(
         db,
@@ -436,6 +437,7 @@ def _agent_runtime_config_dict(model: ModelConfig | None) -> dict:
         "presencePenalty": float(model.presence_penalty) if model is not None and model.presence_penalty is not None else None,
         "frequencyPenalty": float(model.frequency_penalty) if model is not None and model.frequency_penalty is not None else None,
         "maxToken": model.max_token if model is not None and model.max_token is not None else 8192,
+        "streamEnabled": model.stream_enabled if model is not None else 1,
     }
 
 

ai_knowledge_base_v2/apps/backend/app/api/chat.py

@@ -83,7 +83,15 @@ def completions(
     db: Session = Depends(get_db),
     current_user: User = Depends(get_current_user),
 ) -> StreamingResponse:
-    return StreamingResponse(_chat_stream(payload, db, current_user), media_type="text/event-stream")
+    return StreamingResponse(
+        _chat_stream(payload, db, current_user),
+        media_type="text/event-stream",
+        headers={
+            "Cache-Control": "no-cache, no-transform",
+            "Connection": "keep-alive",
+            "X-Accel-Buffering": "no",
+        },
+    )
 
 
 @router.post("/stop")