diff --git a/cuda_core/cuda/core/_launcher.pyi b/cuda_core/cuda/core/_launcher.pyi
index 56f189fe72f..27f7b7842b2 100644
--- a/cuda_core/cuda/core/_launcher.pyi
+++ b/cuda_core/cuda/core/_launcher.pyi
@@ -1,8 +1,11 @@
 # This file was generated by stubgen-pyx v0.2.23 from cuda_core/cuda/core/_launcher.pyx
 
+from collections.abc import Callable
+
 from cuda.core._launch_config import LaunchConfig
 from cuda.core._module import Kernel
 from cuda.core._stream import Stream
+from cuda.core._utils.pycompat import BufferProtocol
 from cuda.core.graph import GraphBuilder
 from cuda.core.typing import IsStreamType
 
@@ -27,7 +30,7 @@ def launch(stream: Stream | GraphBuilder | IsStreamType, config: LaunchConfig, k
         launching kernel.
 
     """
-def host_launch(stream: Stream | GraphBuilder | IsStreamType, fn, *, user_data=None) -> None:
+def host_launch(stream: Stream | GraphBuilder | IsStreamType, fn: Callable[..., object], *, user_data: int | BufferProtocol | None=None) -> None:
     """Enqueue a host function on a stream in stream order.
 
     The function runs on a driver-owned thread once all work enqueued on the
diff --git a/cuda_core/cuda/core/_launcher.pyx b/cuda_core/cuda/core/_launcher.pyx
index f5becd4b316..51397fee524 100644
--- a/cuda_core/cuda/core/_launcher.pyx
+++ b/cuda_core/cuda/core/_launcher.pyx
@@ -22,6 +22,9 @@ from math import prod
 from typing import TYPE_CHECKING
 
 if TYPE_CHECKING:
+    from collections.abc import Callable
+
+    from cuda.core._utils.pycompat import BufferProtocol
     from cuda.core.graph import GraphBuilder
     from cuda.core.typing import IsStreamType
 
@@ -70,7 +73,7 @@ def launch(
         HANDLE_RETURN(cydriver.cuLaunchKernelEx(&drv_cfg, func_handle, args_ptr, NULL))
 
 
-def host_launch(stream: Stream | GraphBuilder | IsStreamType, fn, *, user_data=None) -> None:
+def host_launch(stream: Stream | GraphBuilder | IsStreamType, fn: Callable[..., object], *, user_data: int | BufferProtocol | None = None) -> None:
     """Enqueue a host function on a stream in stream order.
 
     The function runs on a driver-owned thread once all work enqueued on the
diff --git a/cuda_core/cuda/core/graph/_graph_builder.pyi b/cuda_core/cuda/core/graph/_graph_builder.pyi
index bf20e0cb3bf..106a926866f 100644
--- a/cuda_core/cuda/core/graph/_graph_builder.pyi
+++ b/cuda_core/cuda/core/graph/_graph_builder.pyi
@@ -1,10 +1,12 @@
 # This file was generated by stubgen-pyx v0.2.23 from cuda_core/cuda/core/graph/_graph_builder.pyx
 
+from collections.abc import Callable
 from dataclasses import dataclass
 from typing import overload
 
 from cuda.core._stream import Stream
 from cuda.core._utils.cuda_utils import driver
+from cuda.core._utils.pycompat import BufferProtocol
 from cuda.core.graph._graph_definition import GraphCondition, GraphDefinition
 from cuda.core.graph._subclasses import (ChildGraphNode, EventRecordNode,
                                          EventWaitNode,
@@ -408,7 +410,7 @@ class GraphBuilder:
         child : :obj:`~graph.GraphBuilder`
             The child graph builder. Must have finished building.
         """
-    def callback(self, fn, *, user_data=None) -> None:
+    def callback(self, fn: Callable[..., object], *, user_data: int | BufferProtocol | None=None) -> None:
         """Add a host callback to the graph during stream capture.
 
         The callback runs on the host CPU when the graph reaches this point
diff --git a/cuda_core/cuda/core/graph/_graph_builder.pyx b/cuda_core/cuda/core/graph/_graph_builder.pyx
index 2ac92caacbc..194fa98c9c2 100644
--- a/cuda_core/cuda/core/graph/_graph_builder.pyx
+++ b/cuda_core/cuda/core/graph/_graph_builder.pyx
@@ -62,8 +62,11 @@ from cuda.core._utils.cuda_utils import (
 )
 
 if TYPE_CHECKING:
+    from collections.abc import Callable
     from typing import overload
 
+    from cuda.core._utils.pycompat import BufferProtocol
+
     from cuda.core.graph._graph_definition import GraphDefinition
 else:
     # Cython applies a method decorator by rebinding the type's dict entry, so
@@ -947,7 +950,7 @@ cdef class GraphBuilder:
             )
         )
 
-    def callback(self, fn, *, user_data=None) -> None:
+    def callback(self, fn: Callable[..., object], *, user_data: int | BufferProtocol | None = None) -> None:
         """Add a host callback to the graph during stream capture.
 
         The callback runs on the host CPU when the graph reaches this point
diff --git a/cuda_core/cuda/core/graph/_graph_definition.pyi b/cuda_core/cuda/core/graph/_graph_definition.pyi
index 6a139eb7935..871201d4bb0 100644
--- a/cuda_core/cuda/core/graph/_graph_definition.pyi
+++ b/cuda_core/cuda/core/graph/_graph_definition.pyi
@@ -1,12 +1,15 @@
 # This file was generated by stubgen-pyx v0.2.23 from cuda_core/cuda/core/graph/_graph_definition.pyx
 
 """GraphDefinition: explicit CUDA graph definition."""
+from collections.abc import Callable
+
 from cuda.core._device import Device
 from cuda.core._event import Event
 from cuda.core._launch_config import LaunchConfig
 from cuda.core._memory._buffer import Buffer
 from cuda.core._module import Kernel
 from cuda.core._utils.cuda_utils import driver
+from cuda.core._utils.pycompat import BufferProtocol
 from cuda.core.graph._graph_builder import (Graph, GraphCompleteOptions,
                                             GraphDebugPrintOptions)
 from cuda.core.graph._graph_node import GraphNode
@@ -127,7 +130,7 @@ class GraphDefinition:
 
         See :meth:`GraphNode.wait` for full documentation.
         """
-    def callback(self, fn, *, user_data=None) -> HostCallbackNode:
+    def callback(self, fn: Callable[..., object], *, user_data: int | BufferProtocol | None=None) -> HostCallbackNode:
         """Add an entry-point host callback node (no dependencies).
 
         See :meth:`GraphNode.callback` for full documentation.
diff --git a/cuda_core/cuda/core/graph/_graph_definition.pyx b/cuda_core/cuda/core/graph/_graph_definition.pyx
index f9254e8c229..ea97c47bc15 100644
--- a/cuda_core/cuda/core/graph/_graph_definition.pyx
+++ b/cuda_core/cuda/core/graph/_graph_definition.pyx
@@ -31,11 +31,14 @@ from cuda.core._utils.cuda_utils import driver
 from cuda.core.typing import GraphMemoryType
 
 if TYPE_CHECKING:
+    from collections.abc import Callable
+
     from cuda.core._device import Device
     from cuda.core._event import Event
     from cuda.core._launch_config import LaunchConfig
     from cuda.core._memory._buffer import Buffer
     from cuda.core._module import Kernel
+    from cuda.core._utils.pycompat import BufferProtocol
     from cuda.core.graph._graph_builder import (
         Graph,
         GraphCompleteOptions,
@@ -250,7 +253,7 @@ cdef class GraphDefinition:
         """
         return self._entry.wait(event)
 
-    def callback(self, fn, *, user_data=None) -> HostCallbackNode:
+    def callback(self, fn: Callable[..., object], *, user_data: int | BufferProtocol | None = None) -> HostCallbackNode:
         """Add an entry-point host callback node (no dependencies).
 
         See :meth:`GraphNode.callback` for full documentation.
diff --git a/cuda_core/cuda/core/graph/_graph_node.pyi b/cuda_core/cuda/core/graph/_graph_node.pyi
index b22f3d64dcd..a92f09d1467 100644
--- a/cuda_core/cuda/core/graph/_graph_node.pyi
+++ b/cuda_core/cuda/core/graph/_graph_node.pyi
@@ -2,7 +2,7 @@
 
 """GraphNode base class — factory, properties, and builder methods."""
 import weakref
-from collections.abc import Iterable
+from collections.abc import Callable, Iterable
 
 from cuda.core._device import Device
 from cuda.core._event import Event
@@ -10,6 +10,7 @@ from cuda.core._launch_config import LaunchConfig
 from cuda.core._memory._buffer import Buffer
 from cuda.core._module import Kernel
 from cuda.core._utils.cuda_utils import driver
+from cuda.core._utils.pycompat import BufferProtocol
 from cuda.core.graph._adjacency_set_proxy import AdjacencySetProxy
 from cuda.core.graph._graph_definition import GraphCondition, GraphDefinition
 from cuda.core.graph._subclasses import (AllocNode, ChildGraphNode, EmptyNode,
@@ -297,7 +298,7 @@ class GraphNode:
         EventWaitNode
             A new EventWaitNode representing the event wait operation.
         """
-    def callback(self, fn, *, user_data=None) -> object:
+    def callback(self, fn: Callable[..., object], *, user_data: int | BufferProtocol | None=None) -> object:
         """Add a host callback node depending on this node.
 
         The callback runs on the host CPU when the graph reaches this node.
diff --git a/cuda_core/cuda/core/graph/_graph_node.pyx b/cuda_core/cuda/core/graph/_graph_node.pyx
index 4abb1f7106c..d13413a576d 100644
--- a/cuda_core/cuda/core/graph/_graph_node.pyx
+++ b/cuda_core/cuda/core/graph/_graph_node.pyx
@@ -79,7 +79,10 @@ from cuda.core._utils.cuda_utils import driver
 from cuda.core.typing import GraphMemoryType
 
 if TYPE_CHECKING:
+    from collections.abc import Callable
+
     from cuda.core._device import Device
+    from cuda.core._utils.pycompat import BufferProtocol
 
 __all__ = ['GraphNode']
 
@@ -495,7 +498,7 @@ cdef class GraphNode:
         """
         return GN_wait_event(self, event)
 
-    def callback(self, fn, *, user_data=None) -> object:
+    def callback(self, fn: Callable[..., object], *, user_data: int | BufferProtocol | None = None) -> object:
         """Add a host callback node depending on this node.
 
         The callback runs on the host CPU when the graph reaches this node.
diff --git a/cuda_core/examples/host_launch.py b/cuda_core/examples/host_launch.py
index e70ec675b93..0e7bc2d855b 100644
--- a/cuda_core/examples/host_launch.py
+++ b/cuda_core/examples/host_launch.py
@@ -10,7 +10,7 @@
 # ################################################################################
 
 # /// script
-# dependencies = ["cuda_bindings", "cuda_core"]
+# dependencies = ["cuda_core"]
 # ///
 
 import ctypes