diff --git a/msa-ai-service/Dockerfile b/msa-ai-service/Dockerfile
index 63aa625..4b8f6a5 100644
--- a/msa-ai-service/Dockerfile
+++ b/msa-ai-service/Dockerfile
@@ -2,8 +2,17 @@ FROM python:3.10-slim
 
 WORKDIR /app
 
+# SSL 인증서와 필요한 툴 설치
+RUN apt-get update && apt-get install -y --no-install-recommends \
+    ca-certificates \
+    curl \
+    && rm -rf /var/lib/apt/lists/*
+    
 COPY requirements.txt .
 RUN pip install --no-cache-dir -r requirements.txt
+# RUN python -c "from sentence_transformers import SentenceTransformer; \
+#                SentenceTransformer('BM-K/KoSimCSE-roberta').save('./model')"
+
 
 COPY . .
 
diff --git a/msa-ai-service/app/core/config.py b/msa-ai-service/app/core/config.py
index 1885309..6101958 100644
--- a/msa-ai-service/app/core/config.py
+++ b/msa-ai-service/app/core/config.py
@@ -14,5 +14,5 @@
 MONGODB_NAME = os.getenv("MONGODB_NAME", "ai_service_db")
 OPENAI_API_KEY = os.getenv("OPENAI_API_KEY") 
 
-REVIEW_LABELS = ["quantity", "size", "sweet", "salty", "spicy", "deep"]
+REVIEW_LABELS = ["quantity", "size", "sweet", "salty", "spicy", "deep", "sour"]
 POLARITY_LABELS = ["POSITIVE", "NEGATIVE"]
\ No newline at end of file
diff --git a/msa-ai-service/app/ml/embedding_model.py b/msa-ai-service/app/ml/embedding_model.py
new file mode 100644
index 0000000..661dc97
--- /dev/null
+++ b/msa-ai-service/app/ml/embedding_model.py
@@ -0,0 +1 @@
+# Huggingface로 모델 로드하기
\ No newline at end of file
diff --git a/msa-ai-service/app/services/embedding_service.py b/msa-ai-service/app/services/embedding_service.py
index 426c7e6..f3f61c1 100644
--- a/msa-ai-service/app/services/embedding_service.py
+++ b/msa-ai-service/app/services/embedding_service.py
@@ -32,7 +32,7 @@ def embed_and_label_question(question: str):
     질문: {question}
     """
     resp = client.chat.completions.create(
-        model="gpt-4o-mini",
+        model="gpt-41-mini",
         messages=[{"role": "user", "content": prompt}],
         response_format={"type": "json_object"}
     )
@@ -62,7 +62,7 @@ def embed_and_label_review(text: str):
     리뷰: {text}
     """
     resp = client.chat.completions.create(
-        model="gpt-4o-mini",
+        model="gpt-41-mini",
         messages=[{"role": "user", "content": prompt}],
         response_format={"type": "json_object"}
     )
diff --git a/msa-ai-service/app/services/rag_service.py b/msa-ai-service/app/services/rag_service.py
index e9c4e11..f0a26f5 100644
--- a/msa-ai-service/app/services/rag_service.py
+++ b/msa-ai-service/app/services/rag_service.py
@@ -41,9 +41,9 @@ def generate_answer(store_name, menu_name, question, label, reviews):
 1. 반드시 위 리뷰만 사실 근거로 삼아 답변하세요.
 2. {label} 속성에 해당하는 리뷰들 중, 긍정/부정 리뷰 개수를 세어라.
 3. 다음 형식으로 요약하라:
-   - "'{label}' 관련 리뷰 {len(reviews)}건 중 X건은 긍정적이고, Y건은 부정적입니다."
+   - {label} 관련 리뷰 {len(reviews)}건 중 X건은 긍정적이고, Y건은 부정적입니다."
    - 마지막에 결론을 붙여라. (예: "대체로 짜다고 합니다", "의견이 갈립니다", "비율이 비슷합니다", "너무 짜다고 합니다")
-4. 긍정/부정이라는 단어는 쓰지 말고, '{label}' 속성에 맞는 자연스러운 한국어 서술형으로 풀어라.
+4. 긍정/부정이라는 단어는 쓰지 말고, {label} 속성에 맞는 자연스러운 한국어 서술형으로 풀어라.
    - 예: salty → "짜다" / "짜지 않다"
    - 예: quantity → "양이 많다" / "양이 적다"
    - 예: spicy → "맵다" / "안맵다"
@@ -51,7 +51,7 @@ def generate_answer(store_name, menu_name, question, label, reviews):
 """
 
     resp = client.chat.completions.create(
-        model="gpt-4o-mini",
+        model="gpt-41-mini",
         messages=[{"role": "user", "content": prompt}]
     )
     return resp.choices[0].message.content.strip()
diff --git a/msa-ai-service/requirements.txt b/msa-ai-service/requirements.txt
index 2102c31..944f04c 100644
--- a/msa-ai-service/requirements.txt
+++ b/msa-ai-service/requirements.txt
@@ -5,4 +5,8 @@ dnspython
 python-dotenv==1.0.1
 pydantic==2.7.1
 openai
-numpy
\ No newline at end of file
+certifi
+dnspython
+numpy
+sentence-transformers
+torch
\ No newline at end of file