diff --git a/common/log.cpp b/common/log.cpp
index 2d1e74ad1..0f0cb7902 100644
--- a/common/log.cpp
+++ b/common/log.cpp
@@ -438,7 +438,7 @@ void common_log_flush(struct common_log * log) {
     log->resume();
 }
 
-static int common_get_verbosity(enum ggml_log_level level) {
+int common_log_get_verbosity(enum ggml_log_level level) {
     switch (level) {
         case GGML_LOG_LEVEL_DEBUG: return LOG_LEVEL_DEBUG;
         case GGML_LOG_LEVEL_INFO:  return LOG_LEVEL_TRACE;
@@ -452,7 +452,7 @@ static int common_get_verbosity(enum ggml_log_level level) {
 }
 
 void common_log_default_callback(enum ggml_log_level level, const char * text, void * /*user_data*/) {
-    auto verbosity = common_get_verbosity(level);
+    auto verbosity = common_log_get_verbosity(level);
     if (verbosity <= common_log_verbosity_thold) {
         common_log_add(common_log_main(), level, "%s", text);
     }
diff --git a/common/log.h b/common/log.h
index 45d82f4dd..f03358252 100644
--- a/common/log.h
+++ b/common/log.h
@@ -43,6 +43,8 @@ int  common_log_get_verbosity_thold(void);
 
 void common_log_set_verbosity_thold(int verbosity); // not thread-safe
 
+int common_log_get_verbosity(enum ggml_log_level level);
+
 void common_log_default_callback(enum ggml_log_level level, const char * text, void * user_data);
 
 // the common_log uses an internal worker thread to print/write log messages
diff --git a/tests/test-llama-archs.cpp b/tests/test-llama-archs.cpp
index 35a3286e4..6a05c3b96 100644
--- a/tests/test-llama-archs.cpp
+++ b/tests/test-llama-archs.cpp
@@ -65,7 +65,7 @@ static void set_tensor_data(struct ggml_tensor * tensor, void * userdata) {
 }
 
 static void usage(char ** argv) {
-    printf("Usage: %s [-a/--arch arch] [-s/--seed seed] [-o/--out dir] [-v/--verbose] [-h/--help]\n", argv[0]);
+    printf("Usage: %s [-a/--arch arch] [-s/--seed seed] [-o/--out dir] [-v N] [-h/--help]\n", argv[0]);
 }
 
 static std::vector<llama_token> get_tokens(const uint32_t n_tokens, const uint32_t n_vocab, const size_t seed){
@@ -535,22 +535,27 @@ static bool arch_supported(const llm_arch arch) {
     return true;
 }
 
-static int save_models(const llm_arch target_arch, const size_t seed, const ggml_log_level log_level, const std::string & dir) {
+static int save_models(const llm_arch target_arch, const size_t seed, const int verbosity, const std::string & dir) {
     struct user_data_t {
         struct {
             ggml_log_callback callback;
             void * user_data;
-        } original_logger;
-        ggml_log_level min_level; // prints below this log level go to debug log
+        } log_old;
+
+        int verbosity;
+
+        user_data_t(int verbosity) : verbosity(verbosity) {
+            llama_log_get(&log_old.callback, &log_old.user_data);
+        }
     };
-    user_data_t ud;
-    llama_log_get(&ud.original_logger.callback, &ud.original_logger.user_data);
-    ud.min_level = log_level;
+    user_data_t ud(verbosity);
 
     llama_log_set([](ggml_log_level level, const char * text, void * user_data) {
         const user_data_t * ud = (const user_data_t *) user_data;
-        const ggml_log_level level_eff = level >= ud->min_level ? level : GGML_LOG_LEVEL_DEBUG;
-        ud->original_logger.callback(level_eff, text, ud->original_logger.user_data);
+        int verbosity = common_log_get_verbosity(level);
+        if (verbosity <= ud->verbosity) {
+            ud->log_old.callback(level, text, ud->log_old.user_data);
+        }
     }, &ud);
 
     for (const llm_arch & arch : llm_arch_all()) {
@@ -584,26 +589,31 @@ static int save_models(const llm_arch target_arch, const size_t seed, const ggml
             llama_model_save_to_file(model_and_ctx.first.get(), path.c_str());
         }
     }
-    llama_log_set(ud.original_logger.callback, ud.original_logger.user_data);
+    llama_log_set(ud.log_old.callback, ud.log_old.user_data);
     return 0;
 }
 
-static int test_backends(const llm_arch target_arch, const size_t seed, const ggml_log_level log_level) {
+static int test_backends(const llm_arch target_arch, const size_t seed, const int verbosity) {
     struct user_data_t {
         struct {
             ggml_log_callback callback;
             void * user_data;
-        } original_logger;
-        ggml_log_level min_level; // prints below this log level go to debug log
+        } log_old;
+
+        int verbosity;
+
+        user_data_t(int verbosity) : verbosity(verbosity) {
+            llama_log_get(&log_old.callback, &log_old.user_data);
+        }
     };
-    user_data_t ud;
-    llama_log_get(&ud.original_logger.callback, &ud.original_logger.user_data);
-    ud.min_level = log_level;
+    user_data_t ud(verbosity);
 
     llama_log_set([](ggml_log_level level, const char * text, void * user_data) {
         const user_data_t * ud = (const user_data_t *) user_data;
-        const ggml_log_level level_eff = level >= ud->min_level ? level : GGML_LOG_LEVEL_DEBUG;
-        ud->original_logger.callback(level_eff, text, ud->original_logger.user_data);
+        int verbosity = common_log_get_verbosity(level);
+        if (verbosity <= ud->verbosity) {
+            ud->log_old.callback(level, text, ud->log_old.user_data);
+        }
     }, &ud);
 
     const std::vector<llama_token> tokens = get_tokens(128, 128, seed);
@@ -749,20 +759,23 @@ static int test_backends(const llm_arch target_arch, const size_t seed, const gg
             }
         }
     }
-    llama_log_set(ud.original_logger.callback, ud.original_logger.user_data);
+    llama_log_set(ud.log_old.callback, ud.log_old.user_data);
     return all_ok ? 0 : 1;
 }
 
 int main(int argc, char ** argv) {
-    // FIXME these tests are disabled in the CI for macOS-latest-cmake-arm64 because they are segfaulting
+    // init the logger at max verbosity. filter with a custom callback respecting the user-configure verbosity
+    common_log_set_verbosity_thold(LOG_LEVEL_DEBUG);
     common_init();
+
     std::random_device rd;
 
     llm_arch arch = LLM_ARCH_UNKNOWN;
     size_t seed = rd();
-    ggml_log_level log_level = GGML_LOG_LEVEL_ERROR;
     std::string out;
 
+    int verbosity = LOG_LEVEL_ERROR;
+
     for (int i = 1; i < argc; i++) {
         if (strcmp(argv[i], "-h") == 0 || strcmp(argv[i], "--help") == 0) {
             usage(argv);
@@ -789,9 +802,13 @@ int main(int argc, char ** argv) {
                 return 1;
             }
         }
-        if (strcmp(argv[i], "-v") == 0 || strcmp(argv[i], "--verbose") == 0) {
-            log_level = GGML_LOG_LEVEL_INFO;
-            continue;
+        if (strcmp(argv[i], "-v") == 0) {
+            if (i + 1 < argc) {
+                verbosity = std::stoull(argv[++i]);
+            } else {
+                usage(argv);
+                return 1;
+            }
         }
         if (strcmp(argv[i], "-o") == 0 || strcmp(argv[i], "--out") == 0) {
             if (i + 1 < argc) {
@@ -806,9 +823,9 @@ int main(int argc, char ** argv) {
 
     try {
         if (!out.empty()) {
-            return save_models(arch, seed, log_level, out);
+            return save_models(arch, seed, verbosity, out);
         }
-        return test_backends(arch, seed, log_level);
+        return test_backends(arch, seed, verbosity);
     } catch (const std::exception & err) {
         fprintf(stderr, "encountered runtime error: %s\n", err.what());
         return -1;