diff --git a/main.cpp b/main.cpp
index 83816cb..51bb372 100644
--- a/main.cpp
+++ b/main.cpp
@@ -8,6 +8,7 @@
 #include <csignal>
 #include <cstdio>
 #include <cstdlib>
+#include <cstring>
 #include <getopt.h>
 #include <string>
 #include <sys/stat.h>
@@ -17,6 +18,10 @@
 
 static volatile bool g_interrupted = false;
 
+static void noop_log(enum ggml_log_level level, const char * text, void * user_data) {
+  (void)level; (void)text; (void)user_data;
+}
+
 static void sigint_handler(const int signo) {
   (void)signo;
   g_interrupted = true;
@@ -113,13 +118,13 @@ static std::vector<unsigned char> read_clipboard() {
   return data;
 }
 
-static std::string capture_screenshot() {
+static std::string capture_screenshot(bool quiet) {
   const char *tmpdir = getenv("TMPDIR");
   if (!tmpdir)
     tmpdir = "/tmp";
   std::string path = std::string(tmpdir) + "/ocr_screenshot.png";
 
-  fprintf(stderr, "Select a screen region to OCR...\n");
+  if (!quiet) fprintf(stderr, "Select a screen region to OCR...\n");
 
   auto check = [&]() -> bool {
     struct stat st {};
@@ -178,6 +183,7 @@ static void print_usage(const char *prog) {
       "      --json              output in JSON format\n"
       "      --chat-template <name>  chat template (default: auto from model)\n"
       "      --screenshot         interactively select screen region to OCR\n"
+      "  -q, --quiet             suppress info logs, show only OCR result\n"
       "  -h                      show this help\n"
       "\n"
       "If <image_path> is omitted, reads image from clipboard.\n"
@@ -204,6 +210,14 @@ int main(int argc, char **argv) {
   float temp = 0.0f;
   uint32_t seed = LLAMA_DEFAULT_SEED;
   bool json_output = false;
+  bool quiet = false;
+
+  // pre-process -mm to --mmproj so getopt handles it correctly
+  for (int i = 1; i < argc; i++) {
+    if (strcmp(argv[i], "-mm") == 0) {
+      argv[i] = (char *)"--mmproj";
+    }
+  }
 
   while (true) {
     static option long_opts[] = {
@@ -215,10 +229,11 @@ int main(int argc, char **argv) {
         {"model", required_argument, nullptr, 'm'},
         {"chat-template", required_argument, nullptr, 4},
         {"screenshot", no_argument, nullptr, 5},
+        {"quiet", no_argument, nullptr, 6},
         {"help", no_argument, nullptr, 'h'},
         {nullptr, 0, nullptr, 0}};
     int idx = 0;
-    int c = getopt_long(argc, argv, "m:p:t:c:s:h", long_opts, &idx);
+    int c = getopt_long(argc, argv, "m:p:t:c:s:qh", long_opts, &idx);
     if (c == -1)
       break;
     switch (c) {
@@ -240,6 +255,10 @@ int main(int argc, char **argv) {
     case 5:
       screenshot_mode = true;
       break;
+    case 6:
+    case 'q':
+      quiet = true;
+      break;
     case 'm':
       model_path = optarg;
       break;
@@ -270,7 +289,7 @@ int main(int argc, char **argv) {
   }
 
   if (screenshot_mode) {
-    image_path = capture_screenshot();
+    image_path = capture_screenshot(quiet);
     if (image_path.empty()) {
       fprintf(stderr,
               "ERROR: no screenshot tool found. Install gnome-screenshot, "
@@ -287,6 +306,11 @@ int main(int argc, char **argv) {
 
   llama_backend_init();
 
+  if (quiet) {
+    llama_log_set(noop_log, nullptr);
+    mtmd_helper_log_set(noop_log, nullptr);
+  }
+
   llama_model_params mparams = llama_model_default_params();
   mparams.n_gpu_layers = n_gpu_layers;
 
@@ -343,7 +367,7 @@ int main(int argc, char **argv) {
       return 1;
     }
   } else {
-    fprintf(stderr, "Reading image from clipboard...\n");
+    if (!quiet) fprintf(stderr, "Reading image from clipboard...\n");
     auto clip = read_clipboard();
     if (clip.empty()) {
       fprintf(stderr,