diff --git a/app/ai/client.py b/app/ai/client.py
index 45a7401..3ad9a72 100644
--- a/app/ai/client.py
+++ b/app/ai/client.py
@@ -262,14 +262,19 @@ class AIClient:
             data = response.json()
             return data["content"][0]["text"]
         else:
+            payload = {
+                "model": model_config.model,
+                "messages": [{"role": "user", "content": prompt}],
+                "max_tokens": model_config.max_tokens,
+                "chat_template_kwargs": {"enable_thinking": False},
+            }
+            if model_config.temperature is not None:
+                payload["temperature"] = model_config.temperature
+            if model_config.top_p is not None:
+                payload["top_p"] = model_config.top_p
             response = await self._http.post(
                 model_config.endpoint,
-                json={
-                    "model": model_config.model,
-                    "messages": [{"role": "user", "content": prompt}],
-                    "max_tokens": model_config.max_tokens,
-                    "chat_template_kwargs": {"enable_thinking": False},
-                },
+                json=payload,
                 timeout=model_config.timeout,
             )
             response.raise_for_status()
diff --git a/app/core/config.py b/app/core/config.py
index d2f765c..e999900 100644
--- a/app/core/config.py
+++ b/app/core/config.py
@@ -26,6 +26,10 @@ class AIModelConfig(BaseModel):
     # B-0: 4B/26B 에 부여한 실사용 컨텍스트 상한 (char). triage=120k, primary=260k.
     # classify_worker 가 에스컬레이션 판정 시 참고. 0/None 이면 상한 무시.
     context_char_limit: int | None = None
+    # P1 of family-adaptive-bengio (2026-05-23): config-driven sampling profile.
+    # None = MLX/OpenAI server default. Anthropic branch 는 미적용 (별 plan 범위).
+    temperature: float | None = None
+    top_p: float | None = None
 
 
 class DeepSummaryBacklogConfig(BaseModel):
diff --git a/config.yaml b/config.yaml
index 690dd88..716a555 100644
--- a/config.yaml
+++ b/config.yaml
@@ -17,6 +17,7 @@ ai:
       max_tokens: 4096
       timeout: 30
       context_char_limit: 120000
+      temperature: 0.0
 
     # primary: 에스컬레이션 전용. 26B MLX (맥미니 Semaphore(1) 보호 대상).
     primary:
@@ -25,6 +26,8 @@ ai:
       max_tokens: 8192
       timeout: 180
       context_char_limit: 260000
+      temperature: 0.3
+      top_p: 0.9
 
     # fallback: primary 장애 시 최후 방어선. Claude Sonnet 4 API (소액 한도, 자동 trigger).
     # 호출 빈도 낮음 가정 (Mac mini 가 거의 항상 up) → premium 과 budget 공유 OK.