diff --git a/tests/test_vectors.py b/tests/test_vectors.py
index 57b4266..e549589 100644
--- a/tests/test_vectors.py
+++ b/tests/test_vectors.py
@@ -1,6 +1,7 @@
 import uuid
 import turbopuffer as tpuf
 import tests
+import pytest
 from datetime import datetime
 
 
@@ -485,3 +486,12 @@ def test_attribute_types():
         ]],
     )
     assert len(results) == 1
+
+def test_not_found_error():
+    ns = tpuf.Namespace(tests.test_prefix + 'not_found')
+
+    with pytest.raises(tpuf.NotFoundError):
+        ns.query(
+            top_k=5,
+            vector=[0.0, 0.0],
+        )
diff --git a/turbopuffer/__init__.py b/turbopuffer/__init__.py
index d5a0bf0..4ff5a3f 100644
--- a/turbopuffer/__init__.py
+++ b/turbopuffer/__init__.py
@@ -28,4 +28,4 @@ def dump_json_bytes(obj): return json.dumps(obj, cls=NumpyEncoder).encode()
 from turbopuffer.namespace import Namespace, namespaces
 from turbopuffer.vectors import VectorColumns, VectorRow, VectorResult
 from turbopuffer.query import VectorQuery, Filters
-from turbopuffer.error import TurbopufferError, AuthenticationError, APIError
+from turbopuffer.error import TurbopufferError, AuthenticationError, APIError, NotFoundError
diff --git a/turbopuffer/backend.py b/turbopuffer/backend.py
index b5b2992..a7c9660 100644
--- a/turbopuffer/backend.py
+++ b/turbopuffer/backend.py
@@ -5,7 +5,7 @@
 import requests
 import turbopuffer as tpuf
 import gzip
-from turbopuffer.error import AuthenticationError, APIError
+from turbopuffer.error import AuthenticationError, raise_api_error
 from typing import Optional, List
 
 
@@ -121,7 +121,7 @@ def make_api_request(self,
                     try:
                         content = response.json()
                     except json.JSONDecodeError as err:
-                        raise APIError(response.status_code, traceback.format_exception_only(err), response.text)
+                        raise_api_error(response.status_code, traceback.format_exception_only(err), response.text)
 
                     if response.ok:
                         performance['total_time'] = time.monotonic() - start
@@ -130,9 +130,9 @@ def make_api_request(self,
                             'performance': performance,
                         })
                     else:
-                        raise APIError(response.status_code, content.get('status', 'error'), content.get('error', ''))
+                        raise_api_error(response.status_code, content.get('status', 'error'), content.get('error', ''))
                 else:
-                    raise APIError(response.status_code, 'Server returned non-JSON response', response.text)
+                    raise_api_error(response.status_code, 'Server returned non-JSON response', response.text)
             except requests.HTTPError as http_err:
                 retry_attempt += 1
                 # print(traceback.format_exc())
@@ -141,6 +141,6 @@ def make_api_request(self,
                     time.sleep(2 ** retry_attempt)  # exponential falloff up to 64 seconds for 6 retries.
                 else:
                     print(f'Request failed after {retry_attempt} attempts...')
-                    raise APIError(http_err.response.status_code,
+                    raise_api_error(http_err.response.status_code,
                                    f'Request to {http_err.request.url} failed after {retry_attempt} attempts',
                                    str(http_err))
diff --git a/turbopuffer/error.py b/turbopuffer/error.py
index 51760bd..1bcb38c 100644
--- a/turbopuffer/error.py
+++ b/turbopuffer/error.py
@@ -11,3 +11,12 @@ def __init__(self, status_code: int, status_name: str, message: str):
         self.status_code = status_code
         self.status_name = status_name
         super().__init__(f'{status_name} (HTTP {status_code}): {message}')
+
+class NotFoundError(APIError):
+    pass
+
+def raise_api_error(status_code, status_name, message):
+    if status_code == 404:
+        raise NotFoundError(status_code, status_name, message)
+    else:
+        raise APIError(status_code, status_name, message)