diff --git a/ramalama/cli.py b/ramalama/cli.py
index e6afd28e..69f404d0 100644
--- a/ramalama/cli.py
+++ b/ramalama/cli.py
@@ -663,6 +663,8 @@ def _get_source(args):
     if smodel.type == "OCI":
         return src
     else:
+        if not smodel.exists(args):
+            return smodel.pull(args)
         return smodel.path(args)
 
 
diff --git a/ramalama/common.py b/ramalama/common.py
index 7f4f2e4e..fdfa1aad 100644
--- a/ramalama/common.py
+++ b/ramalama/common.py
@@ -1,19 +1,29 @@
 """ramalama common module."""
-
 import glob
 import hashlib
 import os
 import random
+import logging
 import shutil
 import string
 import subprocess
+import time
 import sys
 import urllib.request
 import urllib.error
+import ramalama.console as console
+
 from ramalama.http_client import HttpClient
 
+
+logging.basicConfig(
+    level=logging.WARNING,
+    format="%(asctime)s - %(levelname)s - %(message)s"
+)
+
 MNT_DIR = "/mnt/models"
 MNT_FILE = f"{MNT_DIR}/model.file"
+HTTP_RANGE_NOT_SATISFIABLE = 416
 
 
 def container_manager():
@@ -165,25 +175,61 @@ def download_file(url, dest_path, headers=None, show_progress=True):
         headers (dict): Optional headers to include in the request.
         show_progress (bool): Whether to show a progress bar during download.
 
-    Returns:
-        None
+    Raises:
+        RuntimeError: If the download fails after multiple attempts.
     """
-    http_client = HttpClient()
-
     headers = headers or {}
 
-    # if we are not a tty, don't show progress, can pollute CI output and such
+    # If not running in a TTY, disable progress to prevent CI pollution
     if not sys.stdout.isatty():
         show_progress = False
 
-    try:
-        http_client.init(url=url, headers=headers, output_file=dest_path, progress=show_progress)
-    except urllib.error.HTTPError as e:
-        if e.code == 416:  # Range not satisfiable
-            if show_progress:
-                print(f"File {url} already fully downloaded.")
-        else:
-            raise e
+    http_client = HttpClient()
+    max_retries = 5  # Stop after 5 failures
+    retries = 0
+
+    while retries < max_retries:
+        try:
+            # Initialize HTTP client for the request
+            http_client.init(url=url, headers=headers, output_file=dest_path, progress=show_progress)
+            return  # Exit function if successful
+
+        except urllib.error.HTTPError as e:
+            if e.code == HTTP_RANGE_NOT_SATISFIABLE:  # "Range Not Satisfiable" error (file already downloaded)
+                return  # No need to retry
+
+        except urllib.error.URLError as e:
+            console.error(f"Network Error: {e.reason}")
+            retries += 1
+
+        except TimeoutError:
+            retries += 1
+            console.warning(f"TimeoutError: The server took too long to respond. Retrying {retries}/{max_retries}...")
+
+        except RuntimeError as e:  # Catch network-related errors from HttpClient
+            retries += 1
+            console.warning(f"{e}. Retrying {retries}/{max_retries}...")
+
+        except IOError as e:
+            retries += 1
+            console.warning(f"I/O Error: {e}. Retrying {retries}/{max_retries}...")
+
+        except Exception as e:
+            console.error(f"Unexpected error: {str(e)}")
+            raise
+
+        if retries >= max_retries:
+            error_message = (
+                "\nDownload failed after multiple attempts.\n"
+                "Possible causes:\n"
+                "- Internet connection issue\n"
+                "- Server is down or unresponsive\n"
+                "- Firewall or proxy blocking the request\n"
+            )
+            console.error(error_message)
+            sys.exit(1)
+
+        time.sleep(2 ** retries * 0.1)  # Exponential backoff (0.1s, 0.2s, 0.4s...)
 
 
 def engine_version(engine):
diff --git a/ramalama/console.py b/ramalama/console.py
new file mode 100644
index 00000000..f7980c7a
--- /dev/null
+++ b/ramalama/console.py
@@ -0,0 +1,35 @@
+import os
+import sys
+import locale
+import logging
+
+def is_locale_utf8():
+    """Check if the system locale is UTF-8."""
+    return 'UTF-8' in os.getenv('LC_CTYPE', '') or 'UTF-8' in os.getenv('LANG', '')
+
+def supports_emoji():
+    """Detect if the terminal supports emoji output."""
+    term = os.getenv("TERM")
+    if not term or term in ("dumb", "linux"):
+        return False
+
+    return is_locale_utf8()
+
+# Allow users to override emoji support via an environment variable
+# If RAMALAMA_FORCE_EMOJI is not set, it defaults to checking supports_emoji()
+RAMALAMA_FORCE_EMOJI = os.getenv("RAMALAMA_FORCE_EMOJI")
+FORCE_EMOJI = RAMALAMA_FORCE_EMOJI.lower() == "true" if RAMALAMA_FORCE_EMOJI else None
+EMOJI = FORCE_EMOJI if FORCE_EMOJI is not None else supports_emoji()
+
+# Define emoji-aware logging messages
+def error(msg):
+    formatted_msg = f"❌ {msg}" if EMOJI else f"[ERROR] {msg}"
+    logging.error(formatted_msg)
+
+def warning(msg):
+    formatted_msg = f"⚠️  {msg}" if EMOJI else f"[WARNING] {msg}"
+    logging.warning(formatted_msg)
+
+def info(msg):
+    formatted_msg = f"ℹ️  {msg}" if EMOJI else f"[INFO] {msg}"
+    logging.info(formatted_msg)
diff --git a/ramalama/ollama.py b/ramalama/ollama.py
index 32aa4565..4432cb4d 100644
--- a/ramalama/ollama.py
+++ b/ramalama/ollama.py
@@ -107,6 +107,8 @@ def pull(self, args):
         try:
             return init_pull(repos, accept, registry_head, model_name, model_tag, models, model_path, self.model)
         except urllib.error.HTTPError as e:
+            if "Not Found" in e.reason:
+                raise KeyError(f"{self.model} was not found in the Ollama registry")
             raise KeyError(f"failed to pull {registry_head}: " + str(e).strip("'"))
 
     def model_path(self, args):
diff --git a/test/system/040-serve.bats b/test/system/040-serve.bats
index f3680703..311e9c26 100755
--- a/test/system/040-serve.bats
+++ b/test/system/040-serve.bats
@@ -45,7 +45,7 @@ verify_begin=".*run --rm -i --label RAMALAMA --security-opt=label=disable --name
      fi
 
     run_ramalama 1 serve MODEL
-    is "$output" ".*Error: failed to pull .*MODEL" "failed to pull model"
+    is "$output" "Error: MODEL was not found in the Ollama registry"
 }
 
 @test "ramalama --detach serve" {
diff --git a/test/system/050-pull.bats b/test/system/050-pull.bats
index 30299780..8559f32c 100644
--- a/test/system/050-pull.bats
+++ b/test/system/050-pull.bats
@@ -26,7 +26,7 @@ load setup_suite
 
     random_image_name=i_$(safename)
     run_ramalama 1 pull ${random_image_name}
-    is "$output" "Error: failed to pull https://registry.ollama.ai/v2/library/${random_image_name}: HTTP Error 404: Not Found" "image does not exist"
+    is "$output" "Error: ${random_image_name} was not found in the Ollama registry"
 }
 
 # bats test_tags=distro-integration
diff --git a/test/system/055-convert.bats b/test/system/055-convert.bats
index c6afdcab..7a331b20 100644
--- a/test/system/055-convert.bats
+++ b/test/system/055-convert.bats
@@ -10,7 +10,7 @@ load helpers
     run_ramalama 2 convert tiny
     is "$output" ".*ramalama convert: error: the following arguments are required: TARGET"
     run_ramalama 1 convert bogus foobar
-    is "$output" "Error: bogus does not exist"
+    is "$output" "Error: bogus was not found in the Ollama registry"
 }
 
 @test "ramalama convert file to image" {