diff --git a/.release-please-manifest.json b/.release-please-manifest.json
index fd599489..4540b55c 100644
--- a/.release-please-manifest.json
+++ b/.release-please-manifest.json
@@ -1,3 +1,3 @@
 {
-  ".": "0.2.0-alpha.63"
+  ".": "0.2.0-alpha.64"
 }
\ No newline at end of file
diff --git a/CHANGELOG.md b/CHANGELOG.md
index d322990b..8702125e 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -5,6 +5,19 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/)
 and this project adheres to [Semantic Versioning](http://semver.org/spec/v2.0.0.html).
 
+## 0.2.0-alpha.64 (2025-06-13)
+
+Full Changelog: [v0.2.0-alpha.63...v0.2.0-alpha.64](https://github.com/openlayer-ai/openlayer-python/compare/v0.2.0-alpha.63...v0.2.0-alpha.64)
+
+### Bug Fixes
+
+* **client:** correctly parse binary response | stream ([698f29f](https://github.com/openlayer-ai/openlayer-python/commit/698f29f07c95c7bc121b01cd89500699d6f163e0))
+
+
+### Chores
+
+* **tests:** run tests in parallel ([74b43e2](https://github.com/openlayer-ai/openlayer-python/commit/74b43e268a5f7c96bb1139e0e9d5f555cc5f4b13))
+
 ## 0.2.0-alpha.63 (2025-06-03)
 
 Full Changelog: [v0.2.0-alpha.62...v0.2.0-alpha.63](https://github.com/openlayer-ai/openlayer-python/compare/v0.2.0-alpha.62...v0.2.0-alpha.63)
diff --git a/pyproject.toml b/pyproject.toml
index 99b45518..1f2bed0b 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -1,6 +1,6 @@
 [project]
 name = "openlayer"
-version = "0.2.0-alpha.63"
+version = "0.2.0-alpha.64"
 description = "The official Python library for the openlayer API"
 dynamic = ["readme"]
 license = "Apache-2.0"
@@ -60,6 +60,7 @@ dev-dependencies = [
     "importlib-metadata>=6.7.0",
     "rich>=13.7.1",
     "nest_asyncio==1.6.0",
+    "pytest-xdist>=3.6.1",
 ]
 
 [tool.rye.scripts]
@@ -131,7 +132,7 @@ replacement = '[\1](https://github.com/openlayer-ai/openlayer-python/tree/main/\
 
 [tool.pytest.ini_options]
 testpaths = ["tests"]
-addopts = "--tb=short"
+addopts = "--tb=short -n auto"
 xfail_strict = true
 asyncio_mode = "auto"
 asyncio_default_fixture_loop_scope = "session"
diff --git a/requirements-dev.lock b/requirements-dev.lock
index 0524201f..1b2318b0 100644
--- a/requirements-dev.lock
+++ b/requirements-dev.lock
@@ -33,6 +33,8 @@ distro==1.8.0
 exceptiongroup==1.2.2
     # via anyio
     # via pytest
+execnet==2.1.1
+    # via pytest-xdist
 filelock==3.12.4
     # via virtualenv
 h11==0.14.0
@@ -84,7 +86,9 @@ pygments==2.18.0
 pyright==1.1.399
 pytest==8.3.3
     # via pytest-asyncio
+    # via pytest-xdist
 pytest-asyncio==0.24.0
+pytest-xdist==3.7.0
 python-dateutil==2.8.2
     # via pandas
     # via time-machine
diff --git a/src/openlayer/_base_client.py b/src/openlayer/_base_client.py
index 718469f7..b8a466eb 100644
--- a/src/openlayer/_base_client.py
+++ b/src/openlayer/_base_client.py
@@ -1071,7 +1071,14 @@ def _process_response(
     ) -> ResponseT:
         origin = get_origin(cast_to) or cast_to
 
-        if inspect.isclass(origin) and issubclass(origin, BaseAPIResponse):
+        if (
+            inspect.isclass(origin)
+            and issubclass(origin, BaseAPIResponse)
+            # we only want to actually return the custom BaseAPIResponse class if we're
+            # returning the raw response, or if we're not streaming SSE, as if we're streaming
+            # SSE then `cast_to` doesn't actively reflect the type we need to parse into
+            and (not stream or bool(response.request.headers.get(RAW_RESPONSE_HEADER)))
+        ):
             if not issubclass(origin, APIResponse):
                 raise TypeError(f"API Response types must subclass {APIResponse}; Received {origin}")
 
@@ -1574,7 +1581,14 @@ async def _process_response(
     ) -> ResponseT:
         origin = get_origin(cast_to) or cast_to
 
-        if inspect.isclass(origin) and issubclass(origin, BaseAPIResponse):
+        if (
+            inspect.isclass(origin)
+            and issubclass(origin, BaseAPIResponse)
+            # we only want to actually return the custom BaseAPIResponse class if we're
+            # returning the raw response, or if we're not streaming SSE, as if we're streaming
+            # SSE then `cast_to` doesn't actively reflect the type we need to parse into
+            and (not stream or bool(response.request.headers.get(RAW_RESPONSE_HEADER)))
+        ):
             if not issubclass(origin, AsyncAPIResponse):
                 raise TypeError(f"API Response types must subclass {AsyncAPIResponse}; Received {origin}")
 
diff --git a/src/openlayer/_version.py b/src/openlayer/_version.py
index 37ec914f..c23fffab 100644
--- a/src/openlayer/_version.py
+++ b/src/openlayer/_version.py
@@ -1,4 +1,4 @@
 # File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
 
 __title__ = "openlayer"
-__version__ = "0.2.0-alpha.63"  # x-release-please-version
+__version__ = "0.2.0-alpha.64"  # x-release-please-version