diff --git a/src/cpp/server/ollama_api.cpp b/src/cpp/server/ollama_api.cpp
index 7687caab4..420306c56 100644
--- a/src/cpp/server/ollama_api.cpp
+++ b/src/cpp/server/ollama_api.cpp
@@ -1,5 +1,6 @@
 #include "lemon/ollama_api.h"
 #include "lemon/model_types.h"
+#include "lemon/runtime_config.h"
 #include <iostream>
 #include <lemon/utils/aixlog.hpp>
 #include <sstream>
@@ -1181,6 +1182,13 @@ void OllamaApi::handle_pull(const httplib::Request& req, httplib::Response& res)
             res.set_content(error.dump(), "application/json");
             return;
         }
+        auto* cfg = RuntimeConfig::global();
+        if (cfg && cfg->offline()) {
+            res.status = 400;
+            json error = {{"error", "Lemond is in offline mode, models not downloaded"}, {"code", "lemond_offline"}};
+            res.set_content(error.dump(), "application/json");
+            return;
+        }
 
         LOG(INFO, "OllamaApi") << "POST /api/pull - Pulling model: " << name << std::endl;
 
diff --git a/src/cpp/server/server.cpp b/src/cpp/server/server.cpp
index b2b0327ab..c63ca0c50 100644
--- a/src/cpp/server/server.cpp
+++ b/src/cpp/server/server.cpp
@@ -3548,6 +3548,13 @@ void Server::handle_pull(const httplib::Request& req, httplib::Response& res) {
             return;
         }
 
+        if (config_->offline()) {
+            res.status = 400;
+            nlohmann::json error = {{"error", "Lemond is in offline mode, models not downloaded"}, {"code", "lemond_offline"}};
+            res.set_content(error.dump(), "application/json");
+            return;
+        }
+
         if (stream) {
             auto operation = [this, model_name, request_json, do_not_upgrade](DownloadProgressCallback progress_cb) {
                 model_manager_->download_model(model_name, request_json, do_not_upgrade, progress_cb);