diff --git a/.env.example b/.env.example
index 41374cd..08dc15e 100644
--- a/.env.example
+++ b/.env.example
@@ -365,6 +365,7 @@ OLLAMA_111_MODEL_FALLBACK=llama3.2:latest
 OLLAMA_111_MODEL_DOWNGRADE_PATTERNS=qwen3:*,deepseek-r1:*,hermes3:*,llama3.1:*,qwen2.5:*,qwen2.5-coder:*,gemma3:*,minicpm-v:*,llava:*,*:7b*,*:8b*,*:14b*,*:32b*,*:70b*
 OLLAMA_111_KEEP_ALIVE=5m
 OLLAMA_111_MAX_TIMEOUT=45
+OLLAMA_111_NUM_CTX=4096
 
 # [預設 true] OpenClaw Q&A 先走 Ollama，品質不足或失敗時才 fallback Gemini/NIM
 # 主機不提供單 caller override；一律走 OLLAMA_HOST_PRIMARY → OLLAMA_HOST_SECONDARY → OLLAMA_HOST_FALLBACK
diff --git a/config.py b/config.py
index ed44c8e..0a3212f 100644
--- a/config.py
+++ b/config.py
@@ -323,7 +323,7 @@ YOUTUBE_API_KEY = os.getenv('YOUTUBE_API_KEY', '')
 # ==========================================
 # 系統版本與路徑
 # ==========================================
-SYSTEM_VERSION = "V10.363"
+SYSTEM_VERSION = "V10.364"
 LOG_FILE_PATH = os.path.join(BASE_DIR, 'logs/system.log')
 public_url = PUBLIC_URL  # 用於模板顯示
 
diff --git a/docs/AI_INTELLIGENCE_MODULE_SOT.md b/docs/AI_INTELLIGENCE_MODULE_SOT.md
index 57d6ab6..eb4413c 100644
--- a/docs/AI_INTELLIGENCE_MODULE_SOT.md
+++ b/docs/AI_INTELLIGENCE_MODULE_SOT.md
@@ -2,7 +2,7 @@
 
 > **最後更新**: 2026-05-21 (台北時間)
 > **狀態**: 🟢 四 AI Agent 自動化閉環已落地；LLM 路由紅線升級為 Ollama-first 三主機級聯，Gemini 備援預設關閉
-> **適用版本**: V10.362
+> **適用版本**: V10.364
 
 ---
 
@@ -25,7 +25,7 @@
 - OpenClaw 週報、月報、Meta analysis、日報洞察、Telegram PPT 分析與 MCP fallback 也必須 Ollama-first；Gemini caller 只能帶 `_gemini_fallback` 或明確 fallback caller 語意，且不得先於 Ollama/NIM 被呼叫。
 - OpenClaw 週報、月報、Meta analysis、日報洞察與每日報告的 Gemini/NIM 備援 caller 必須登錄在 caller registry、AI 觀測台 agent group 與 Telegram 狀態統計，避免 fallback 用量被歸類為未知或漏算。
 - Gemini API 出站有第二道 kill switch：`GEMINI_FALLBACK_ENABLED` 預設為 `false`。即使 `GEMINI_API_KEY` 存在，通用 AI fallback、OpenClaw 報告/QA/PPT/圖片、MCP Grounding 與 Code Review L3 都不得呼叫 Gemini；只有操作員明確設為 `true` 時，Gemini 才能作緊急備援。
-- 111 `192.168.0.111` 只是最後一道 Mac fallback，不承接 7B+、vision、long-context 模型長駐；`OllamaService.generate()` 落到 111 時會將 `qwen3`、`deepseek-r1`、`hermes3`、`qwen2.5*`、`gemma3`、`llava`、`minicpm-v` 與 7B+ 模型依 `OLLAMA_111_MODEL_DOWNGRADE_PATTERNS` 降級到 `OLLAMA_111_MODEL_FALLBACK=llama3.2:latest`，並以 `OLLAMA_111_KEEP_ALIVE=5m`、`OLLAMA_111_MAX_TIMEOUT=45` 封頂，避免 16GB RAM 主機被大 context runner 與 24h keep-alive 壓到 swap。
+- 111 `192.168.0.111` 只是最後一道 Mac fallback，不承接 7B+、vision、long-context 模型長駐；`OllamaService.generate()` 落到 111 時會將 `qwen3`、`deepseek-r1`、`hermes3`、`qwen2.5*`、`gemma3`、`llava`、`minicpm-v` 與 7B+ 模型依 `OLLAMA_111_MODEL_DOWNGRADE_PATTERNS` 降級到 `OLLAMA_111_MODEL_FALLBACK=llama3.2:latest`，並以 `OLLAMA_111_KEEP_ALIVE=5m`、`OLLAMA_111_MAX_TIMEOUT=45`、`OLLAMA_111_NUM_CTX=4096` 封頂，避免 16GB RAM 主機被大 context runner 與 24h keep-alive 壓到 swap。
 
 ## 一、四 AI Agent 路由架構
 
diff --git a/docs/memory/history_logs.md b/docs/memory/history_logs.md
index 31bd1ed..4e51660 100644
--- a/docs/memory/history_logs.md
+++ b/docs/memory/history_logs.md
@@ -13,6 +13,7 @@
 ## 📅 詳細更新日誌 (考古存檔)
 
 ### 2026-05-21：瀏覽器測試守門與 PChome 熱路徑優化
+- **V10.364 111 context cap**: 111 fallback 即使降到 `llama3.2:latest`，Ollama 仍可能用 131k context 啟動 runner，導致 3B 模型也吃到 10GB+；新增 `OLLAMA_111_NUM_CTX=4096`，落到 111 時強制縮 context，並把 `llama3.2:latest` 加入零成本模型表，避免觀測台 unknown model warning。
 - **V10.363 Dashing Diva variant-safe search**: PChome/MOMO matcher 針對 Dashing Diva 美甲片補「商品頁目錄有 30片/盒、MOMO 標題省略片數」的安全豁免，只限同品牌、同美甲片線、同具名款式錨點；搜尋詞也優先帶出 `月影柔霧`、`銀絲柔彩` 等款式名，降低同系列不同款式互撞。
 - **V10.362 111 fallback shrink-to-3B**: 111 Mac 實測 `hermes3` / `qwen2.5-coder` 雖是 7B/8B，但 large context runner 仍會佔用 6-10GB RSS 並推高 swap；111 fallback 改為所有 7B+、vision 與 long-context 文字生成都降級到 `llama3.2:latest`，`ai_calls.model` 也會記錄實際降級模型並把原請求模型放入 `meta.requested_model`。
 - **V10.361 111 fallback resource guard**: 實測 111 Mac 高 load 主要來自 Codex app / WindowServer 前台負載，且 Ollama 曾因 fallback 載入 `qwen3:14b` 造成 16GB RAM / swap 壓力；已手動 unload 111 上的重模型，並讓 `OllamaService.generate()` 落到 111 時自動把 14B+ 模型降到 `OLLAMA_111_MODEL_FALLBACK`、`keep_alive` 縮至 `OLLAMA_111_KEEP_ALIVE=5m`、timeout 封頂 `OLLAMA_111_MAX_TIMEOUT=45`。GCP-A/GCP-B 仍可跑 `qwen3:14b`，111 只做短時最後備援。
diff --git a/services/ai_call_logger.py b/services/ai_call_logger.py
index dc05b6e..3dff160 100644
--- a/services/ai_call_logger.py
+++ b/services/ai_call_logger.py
@@ -66,6 +66,7 @@ COST_TABLE: Dict[str, Dict[str, float]] = {
     'minicpm-v:latest':     {'in': 0.0, 'out': 0.0},  # Phase 14 PPT vision（5.5GB）
     'llava:latest':         {'in': 0.0, 'out': 0.0},  # 2026-05-04 已拉（4.7GB），Vision 備援
     'llama3.1:8b':          {'in': 0.0, 'out': 0.0},
+    'llama3.2:latest':      {'in': 0.0, 'out': 0.0},  # 111 final fallback 3B
     'bge-m3:latest':        {'in': 0.0, 'out': 0.0},
 }
 
diff --git a/services/ollama_service.py b/services/ollama_service.py
index 74dc5fe..d9f1157 100644
--- a/services/ollama_service.py
+++ b/services/ollama_service.py
@@ -58,6 +58,7 @@ COPY_TIMEOUT = int(os.getenv('OLLAMA_COPY_TIMEOUT', '180'))  # 文案生成專
 EMBED_TIMEOUT = int(os.getenv('OLLAMA_EMBED_TIMEOUT', os.getenv('EMBEDDING_TIMEOUT', '45')))
 FALLBACK_111_KEEP_ALIVE = os.getenv('OLLAMA_111_KEEP_ALIVE', '5m')
 FALLBACK_111_MAX_TIMEOUT = int(os.getenv('OLLAMA_111_MAX_TIMEOUT', '45'))
+FALLBACK_111_NUM_CTX = int(os.getenv('OLLAMA_111_NUM_CTX', '4096'))
 FALLBACK_111_MODEL = os.getenv('OLLAMA_111_MODEL_FALLBACK', 'llama3.2:latest')
 FALLBACK_111_MODEL_PATTERNS = tuple(
     pattern.strip().lower()
@@ -140,6 +141,15 @@ def _effective_timeout_for_host(timeout_s: int, host: str) -> int:
     return timeout_s
 
 
+def _cap_111_options(options: Dict[str, Any]) -> None:
+    """111 fallback 強制縮 context，避免 3B/7B 仍因 131k context 吃爆記憶體。"""
+    try:
+        requested_num_ctx = int(options.get("num_ctx") or FALLBACK_111_NUM_CTX)
+    except (TypeError, ValueError):
+        requested_num_ctx = FALLBACK_111_NUM_CTX
+    options["num_ctx"] = min(requested_num_ctx, FALLBACK_111_NUM_CTX)
+
+
 def _canonical_host_chain() -> List[str]:
     """Return the approved static fallback chain without duplicates."""
     chain: List[str] = []
@@ -412,6 +422,7 @@ class OllamaService:
             payload["model"] = effective_model
             if _is_111_fallback_host(current_host):
                 payload["keep_alive"] = FALLBACK_111_KEEP_ALIVE
+                _cap_111_options(payload["options"])
             elif keep_alive:
                 payload["keep_alive"] = keep_alive
 
diff --git a/tests/test_ollama_resolve.py b/tests/test_ollama_resolve.py
index aeba614..a2a0cc7 100644
--- a/tests/test_ollama_resolve.py
+++ b/tests/test_ollama_resolve.py
@@ -244,6 +244,7 @@ def test_111_fallback_downgrades_heavy_model_and_shortens_keep_alive(monkeypatch
     monkeypatch.setattr(oss, "FALLBACK_111_MODEL", "qwen2.5:7b-instruct")
     monkeypatch.setattr(oss, "FALLBACK_111_KEEP_ALIVE", "5m")
     monkeypatch.setattr(oss, "FALLBACK_111_MAX_TIMEOUT", 45)
+    monkeypatch.setattr(oss, "FALLBACK_111_NUM_CTX", 4096)
     monkeypatch.setattr(oss, "FALLBACK_111_MODEL_PATTERNS", ("qwen3:14b",))
 
     fake_resp = MagicMock(status_code=200)
@@ -261,6 +262,7 @@ def test_111_fallback_downgrades_heavy_model_and_shortens_keep_alive(monkeypatch
     payload = mock_post.call_args.kwargs["json"]
     assert payload["model"] == "qwen2.5:7b-instruct"
     assert payload["keep_alive"] == "5m"
+    assert payload["options"]["num_ctx"] == 4096
     assert mock_post.call_args.kwargs["timeout"] == 45
     assert resp.model == "qwen2.5:7b-instruct"
 
@@ -271,6 +273,7 @@ def test_111_fallback_keeps_light_model_but_caps_timeout(monkeypatch):
 
     monkeypatch.setattr(oss, "FALLBACK_111_KEEP_ALIVE", "5m")
     monkeypatch.setattr(oss, "FALLBACK_111_MAX_TIMEOUT", 45)
+    monkeypatch.setattr(oss, "FALLBACK_111_NUM_CTX", 4096)
     svc = oss.OllamaService(host="http://192.168.0.111:11434", model="llama3.2:latest")
 
     with patch("services.ollama_service.requests.post", side_effect=Timeout):
@@ -286,6 +289,7 @@ def test_111_fallback_downgrades_hermes_context_heavy_model(monkeypatch):
     monkeypatch.setattr(oss, "FALLBACK_111_MODEL", "llama3.2:latest")
     monkeypatch.setattr(oss, "FALLBACK_111_KEEP_ALIVE", "5m")
     monkeypatch.setattr(oss, "FALLBACK_111_MAX_TIMEOUT", 45)
+    monkeypatch.setattr(oss, "FALLBACK_111_NUM_CTX", 4096)
     monkeypatch.setattr(oss, "FALLBACK_111_MODEL_PATTERNS", ("hermes3:*",))
 
     fake_resp = MagicMock(status_code=200)
@@ -303,4 +307,5 @@ def test_111_fallback_downgrades_hermes_context_heavy_model(monkeypatch):
     payload = mock_post.call_args.kwargs["json"]
     assert payload["model"] == "llama3.2:latest"
     assert payload["keep_alive"] == "5m"
+    assert payload["options"]["num_ctx"] == 4096
     assert resp.model == "llama3.2:latest"