diff --git a/controllers/llamaCPP.cc b/controllers/llamaCPP.cc
index 93a276e46..39696c549 100644
--- a/controllers/llamaCPP.cc
+++ b/controllers/llamaCPP.cc
@@ -1,5 +1,6 @@
 #include "llamaCPP.h"
 #include "llama.h"
+#include "log.h"
 #include "utils/nitro_utils.h"
 
 using namespace inferences;
@@ -441,7 +442,6 @@ bool llamaCPP::loadModelImpl(const Json::Value &jsonBody) {
         jsonBody.get("cpu_threads", std::thread::hardware_concurrency())
             .asInt();
     params.cont_batching = jsonBody.get("cont_batching", false).asBool();
-
     this->clean_cache_threshold =
         jsonBody.get("clean_cache_threshold", 5).asInt();
     this->caching_enabled = jsonBody.get("caching_enabled", false).asBool();
@@ -451,6 +451,11 @@ bool llamaCPP::loadModelImpl(const Json::Value &jsonBody) {
         jsonBody.get("system_prompt", "ASSISTANT's RULE: ").asString();
     this->pre_prompt = jsonBody.get("pre_prompt", "").asString();
     this->repeat_last_n = jsonBody.get("repeat_last_n", 32).asInt();
+
+    // Set folder for llama log
+    std::string llama_log_folder =
+        jsonBody.get("llama_log_folder", "log/").asString();
+    log_set_target(llama_log_folder + "llama.log");
   }
 #ifdef GGML_USE_CUBLAS
   LOG_INFO << "Setting up GGML CUBLAS PARAMS";
diff --git a/controllers/llamaCPP.h b/controllers/llamaCPP.h
index b7b2b27b0..5dc693de5 100644
--- a/controllers/llamaCPP.h
+++ b/controllers/llamaCPP.h
@@ -2486,7 +2486,7 @@ class llamaCPP : public drogon::HttpController<llamaCPP> {
 public:
   llamaCPP() {
     // Some default values for now below
-    // log_disable(); // Disable the log to file feature, reduce bloat for
+    log_enable(); // Disable the log to file feature, reduce bloat for
     // target
     // system ()
     std::vector<std::string> llama_models =