diff --git a/app/dto/model/problem_set.py b/app/dto/model/problem_set.py
index e1501af..14d006a 100644
--- a/app/dto/model/problem_set.py
+++ b/app/dto/model/problem_set.py
@@ -5,9 +5,7 @@
 
 class Selection(BaseModel):
     content: str = Field(description="선택지 내용입니다.")
-    correct: bool = Field(
-        description="정답 여부입니다. 정답이면 True, 오답이면 False입니다."
-    )
+    correct: bool = Field(description="정답 여부입니다. 정답이면 True, 오답이면 False입니다.")
 
 
 class Problem(BaseModel):
diff --git a/app/dto/response/error_response.py b/app/dto/response/error_response.py
new file mode 100644
index 0000000..c92c3e4
--- /dev/null
+++ b/app/dto/response/error_response.py
@@ -0,0 +1,9 @@
+from typing import Literal
+
+from pydantic import BaseModel
+
+
+class ErrorResponse(BaseModel):
+    type: Literal["error"] = "error"
+    code: int
+    message: str
diff --git a/app/dto/response/generate_response.py b/app/dto/response/generate_response.py
index 44bbe61..0500fc8 100644
--- a/app/dto/response/generate_response.py
+++ b/app/dto/response/generate_response.py
@@ -1,11 +1,11 @@
-from typing import List
+from typing import List, Literal
 
 from pydantic import BaseModel
 
 from app.dto.model.problem_set import Selection
 
 
-class ProblemResponse(BaseModel):
+class ProblemDTO(BaseModel):
     number: int
     title: str
     selections: List[Selection]
@@ -13,5 +13,9 @@ class ProblemResponse(BaseModel):
     referencedPages: List[int]
 
 
-class GenerateResponse(BaseModel):
-    quiz: List[ProblemResponse]
+class ProblemSetDTO(BaseModel):
+    quiz: List[ProblemDTO]
+
+
+class GenerateResponse(ProblemSetDTO):
+    type: Literal["quiz"] = "quiz"
diff --git a/app/service/explanation_service.py b/app/service/explanation_service.py
index a25191d..1eec152 100644
--- a/app/service/explanation_service.py
+++ b/app/service/explanation_service.py
@@ -10,7 +10,6 @@ class ExplanationService:
     async def generate_specific_explanation(
         specific_explanation_request: SpecificExplanationRequest,
     ):
-
         title = specific_explanation_request.title
         selections = specific_explanation_request.selections
 
diff --git a/app/service/generate_service.py b/app/service/generate_service.py
index 5a60c76..3f8b8a1 100644
--- a/app/service/generate_service.py
+++ b/app/service/generate_service.py
@@ -8,15 +8,15 @@
 
 import fitz
 import requests
-from fastapi import HTTPException
 from langchain_core.output_parsers import JsonOutputParser
 
 from app.adapter.request_to_gpt import request_to_gpt_returning_text
 from app.dto.model.problem_set import ProblemSet
 from app.dto.request.generate_request import GenerateRequest
+from app.dto.response.error_response import ErrorResponse
 from app.dto.response.generate_response import (
-    GenerateResponse,
-    ProblemResponse,
+    ProblemSetDTO,
+    ProblemDTO, GenerateResponse,
 )
 from app.prompt import prompt_factory
 from app.util.create_chunks import create_page_chunks
@@ -121,7 +121,7 @@ async def generate(generate_request: GenerateRequest):
         try:
             for completed_task in asyncio.as_completed(tasks):
                 try:
-                    result: Optional[GenerateResponse] = await completed_task
+                    result: Optional[ProblemSetDTO] = await completed_task
 
                     if result:
                         for quiz in result.quiz:
@@ -131,21 +131,22 @@ async def generate(generate_request: GenerateRequest):
 
                 except Exception as e:
                     logger.error(f"Task processing failed: {e}")
-                    # 하나가 실패해도 전체 스트림을 끊지 않고 다음 퀴즈 생성을 기다림
-                    continue
+                    status_code = getattr(e, "status_code", 500)
+                    yield ErrorResponse(
+                        code=status_code, message=str(e)
+                    ).model_dump_json() + "\n"
 
         except Exception as e:
             logger.error(f"Critical streaming error: {e}")
-            # 여기서 에러를 던지면 클라이언트(Spring)는 연결이 끊긴 것으로 인식
-            raise HTTPException(status_code=500, detail="Streaming process failed")
+            yield ErrorResponse(code=500, message=str(e)).model_dump_json() + "\n"
 
 
 async def process_single_chunk(
-    gpt_request: dict,
-    parser: JsonOutputParser,
-    referenced_pages: List[int],
-    quiz_type: str,
-) -> Optional[GenerateResponse]:
+        gpt_request: dict,
+        parser: JsonOutputParser,
+        referenced_pages: List[int],
+        quiz_type: str,
+) -> Optional[ProblemSetDTO]:
     with log_elapsed(logger, "request_generate_quiz"):
         try:
             text_response = await request_to_gpt_returning_text(
@@ -180,7 +181,7 @@ async def process_single_chunk(
                     random.shuffle(selections)
 
                 problem_responses.append(
-                    ProblemResponse(
+                    ProblemDTO(
                         number=0,
                         title=q.get("title"),
                         selections=selections,
@@ -194,7 +195,7 @@ async def process_single_chunk(
 
         except Exception as e:
             logger.error(f"Chunk processing error: {e}")
-            return None
+            raise e
 
 
 def _extract_filename(uploaded_url: str) -> str:
diff --git a/app/util/create_chunks.py b/app/util/create_chunks.py
index 26f6517..edff522 100644
--- a/app/util/create_chunks.py
+++ b/app/util/create_chunks.py
@@ -15,7 +15,6 @@ def create_page_chunks(
     total_quiz_count: int,
     max_chunk_count: int,
 ) -> List[ChunkInfo]:
-
     # 청크 별 퀴즈 개수 분배
     chunks: List[ChunkInfo] = []
     for i in range(total_quiz_count):