diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 4561fa5..f8f3a33 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -54,7 +54,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        python-version: ["3.10", "3.13"]
+        python-version: ["3.11", "3.13"]
 
     services:
       mongodb:
@@ -106,7 +106,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        python-version: ["3.10", "3.13"]
+        python-version: ["3.11", "3.13"]
 
     steps:
       - uses: actions/checkout@v7
@@ -137,7 +137,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        python-version: ["3.10", "3.13"]
+        python-version: ["3.11", "3.13"]
         runs-on: [windows-latest, macos-latest]
 
     steps:
diff --git a/docs/tutorials/installation.md b/docs/tutorials/installation.md
index 019933f..52451c8 100644
--- a/docs/tutorials/installation.md
+++ b/docs/tutorials/installation.md
@@ -8,7 +8,7 @@
 :::{tab-item} pip
 :sync: pip
 
-You will need Python 3.10 or later. You can check your version of Python by
+You will need Python 3.11 or later. You can check your version of Python by
 typing into a terminal:
 
 ```sh
diff --git a/docs/tutorials/saving.md b/docs/tutorials/saving.md
index c0a8d7f..178b30b 100644
--- a/docs/tutorials/saving.md
+++ b/docs/tutorials/saving.md
@@ -13,7 +13,7 @@ server dependencies.
 :::{tab-item} pip
 :sync: pip
 
-You will need Python 3.10 or later. You can check your version of Python by
+You will need Python 3.11 or later. You can check your version of Python by
 typing into a terminal:
 
 ```sh
diff --git a/pyproject.toml b/pyproject.toml
index 30d23f6..106307b 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -15,7 +15,7 @@ description = "Integrations for using Tiled with Bluesky"
 readme = { file = "README.md", content-type = "text/markdown" }
 license = "BSD-3-Clause"
 license-files = ["LICENSE"]
-requires-python = ">=3.10"
+requires-python = ">=3.11"
 classifiers = [
   "Development Status :: 1 - Planning",
   "Intended Audience :: Science/Research",
@@ -24,7 +24,6 @@ classifiers = [
   "Programming Language :: Python",
   "Programming Language :: Python :: 3",
   "Programming Language :: Python :: 3 :: Only",
-  "Programming Language :: Python :: 3.10",
   "Programming Language :: Python :: 3.11",
   "Programming Language :: Python :: 3.12",
   "Programming Language :: Python :: 3.13",
@@ -36,13 +35,14 @@ dynamic = ["version"]
 dependencies = [
   "dask",
   "event-model",
+  "httpx2",  # the generation used by the tiled client since 0.2.21
   "mongoquery",
   "ndindex",  # imported by tiled.ndslice
   "numpy",
   "pandas",
   "pyarrow",
   "pytz",
-  "tiled[minimal-client] >=0.2.19",
+  "tiled[minimal-client] >=0.2.21",
   "tzlocal",
   "xarray",
 ]
@@ -88,7 +88,7 @@ test = [
   "redis-json-dict",
   "suitcase-mongo",
   "tifffile",
-  "tiled[server] >=0.2.19",
+  "tiled[server] >=0.2.21",
 ]
 dev = [{ include-group = "test" }]
 docs = [
@@ -133,6 +133,11 @@ filterwarnings = [
   # so they remain readable; zarr warns that these dtypes have no stable Zarr V3
   # specification yet. This warning is emitted by the (in-process) test server.
   "ignore:The data type .* does not have a Zarr V3 specification",
+  # SQLAlchemy 2.1 deprecates the `noload` loader strategy, which tiled's catalog
+  # ORM uses for the Webhook.node relationship. Raising this warning as an error
+  # inside `configure_mappers()` poisons every mapper in the (in-process) test
+  # server; third-party issue, to be fixed in tiled.
+  "ignore:The ``noload`` loader strategy is deprecated:DeprecationWarning",
 ]
 log_level = "INFO"
 testpaths = ["tests"]
@@ -144,7 +149,7 @@ report.exclude_also = ['\.\.\.', 'if typing.TYPE_CHECKING:']
 
 [tool.mypy]
 files = ["src", "tests"]
-python_version = "3.10"
+python_version = "3.11"
 warn_unused_configs = true
 strict = false
 enable_error_code = ["ignore-without-code", "redundant-expr", "truthy-bool"]
@@ -174,7 +179,7 @@ ignore = [
 
 
 [tool.pylint]
-py-version = "3.10"
+py-version = "3.11"
 ignore-paths = [".*/_version.py"]
 reports.output-format = "colorized"
 similarities.ignore-imports = "yes"
diff --git a/src/bluesky_tiled_plugins/clients/bluesky_run.py b/src/bluesky_tiled_plugins/clients/bluesky_run.py
index 7be9ccf..83da862 100644
--- a/src/bluesky_tiled_plugins/clients/bluesky_run.py
+++ b/src/bluesky_tiled_plugins/clients/bluesky_run.py
@@ -1,7 +1,7 @@
 import codecs
 import copy
 import functools
-import httpx
+import httpx2
 import io
 import json
 import keyword
@@ -534,13 +534,13 @@ def validate(
                 msg = "Remote validation failed: " + "; ".join(notes)
                 raise ValidationException(msg, self.item["id"])
 
-        except httpx.HTTPStatusError as e:
+        except httpx2.HTTPStatusError as e:
             # Backcompatibility: if the server does not support validation endpoint,
             # it will return 404 Not Found error; in this case, attempt to validate
             # the data structure locally by the client itself (requires multiple
             # round-trips to the server, but better than nothing).
 
-            if response.status_code == httpx.codes.NOT_FOUND:
+            if response.status_code == httpx2.codes.NOT_FOUND:
                 warnings.warn(
                     "Tiled server does not support remote validation. "
                     "Attempting to validate the data structure by the client.",
diff --git a/src/bluesky_tiled_plugins/writing/tiled_writer.py b/src/bluesky_tiled_plugins/writing/tiled_writer.py
index a95c63e..be50eb0 100644
--- a/src/bluesky_tiled_plugins/writing/tiled_writer.py
+++ b/src/bluesky_tiled_plugins/writing/tiled_writer.py
@@ -9,7 +9,7 @@
 from typing import Any, Optional, cast
 import warnings
 
-import httpx
+import httpx2
 import numpy
 import pyarrow
 from event_model import (
@@ -129,9 +129,7 @@ def concatenate_stream_datums(*docs: StreamDatum):
             "All StreamDatum documents must reference the same stream_resource."
         )
     docs = tuple(sorted(docs, key=lambda doc: doc["indices"]["start"]))
-    for d1, d2 in zip(
-        docs[:-1], docs[1:]
-    ):  # TODO: use itertools.pairwise(docs) in python 3.10+
+    for d1, d2 in itertools.pairwise(docs):
         if d1["indices"]["stop"] != d2["indices"]["start"]:
             raise ValueError("StreamDatum documents must be consecutive.")
 
@@ -908,9 +906,9 @@ def _append_with_retry(self, df_client: DataFrameClient, table, desc_name: str):
         try:
             df_client.append_partition(0, table)
             return
-        except (httpx.TransportError, httpx.HTTPStatusError) as err:
+        except (httpx2.TransportError, httpx2.HTTPStatusError) as err:
             if (
-                isinstance(err, httpx.HTTPStatusError)
+                isinstance(err, httpx2.HTTPStatusError)
                 and err.response.status_code < 500
                 and err.response.status_code != 429
             ):
diff --git a/tests/test_tiled_writer.py b/tests/test_tiled_writer.py
index c2def29..3b53c24 100644
--- a/tests/test_tiled_writer.py
+++ b/tests/test_tiled_writer.py
@@ -6,7 +6,7 @@
 from typing import cast
 from urllib.parse import parse_qs, urlparse
 
-import httpx
+import httpx2
 
 from bluesky import RunEngine
 import bluesky.plan_stubs as bps
@@ -1013,7 +1013,7 @@ def patched_event(name, doc):
     assert lines[6]["name"] == "stop"
 
 
-class _LoseAppendResponseOnce(httpx.BaseTransport):
+class _LoseAppendResponseOnce(httpx2.BaseTransport):
     """Forward the first PATCH /table/partition request to the wrapped transport
     -- so the server fully processes the append -- but raise ReadTimeout to the
     caller, simulating a lost response for an append that actually succeeded."""
@@ -1030,7 +1030,7 @@ def handle_request(self, request):
             and "/table/partition/" in str(request.url)
         ):
             self.tripped = True
-            raise httpx.ReadTimeout("Simulated lost response", request=request)
+            raise httpx2.ReadTimeout("Simulated lost response", request=request)
         return response