diff --git a/pyproject.toml b/pyproject.toml
index d246fe6c..cc39418d 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -142,8 +142,7 @@ disallow_incomplete_defs = false
 module = [
     "docx",
     "fitz",
-    "json_stream",
-    "json_stream.httpx",
+    "ijson",
     "lancedb",
     "param",
     "pptx",
diff --git a/ragna/assistants/_google.py b/ragna/assistants/_google.py
index 24bf9071..1b952a25 100644
--- a/ragna/assistants/_google.py
+++ b/ragna/assistants/_google.py
@@ -1,10 +1,25 @@
-from typing import Iterator
+from typing import AsyncIterator
 
+from ragna._compat import anext
 from ragna.core import PackageRequirement, Requirement, Source
 
 from ._api import ApiAssistant
 
 
+class AsyncIteratorReader:
+    def __init__(self, ait: AsyncIterator[bytes]) -> None:
+        self._ait = ait
+
+    async def read(self, n: int) -> bytes:
+        if n == 0:
+            return b""
+
+        try:
+            return await anext(self._ait)
+        except StopIteration:
+            return b""
+
+
 class GoogleApiAssistant(ApiAssistant):
     _API_KEY_ENV_VAR = "GOOGLE_API_KEY"
     _MODEL: str
@@ -36,15 +51,12 @@ def _instructize_prompt(self, prompt: str, sources: list[Source]) -> str:
             ]
         )
 
-    def _call_api(
+    async def _call_api(
         self, prompt: str, sources: list[Source], *, max_new_tokens: int
-    ) -> Iterator[str]:
-        import json_stream.httpx
+    ) -> AsyncIterator[str]:
+        import ijson
 
-        # TODO: Use the async client and make this method async as soon when json-stream
-        #  supports async JSON stream.
-        #  See https://github.com/daggaz/json-stream/issues/54
-        with self._sync_client.stream(
+        async with self._async_client.stream(
             "POST",
             f"https://generativelanguage.googleapis.com/v1beta/models/{self._MODEL}:streamGenerateContent",
             params={"key": self._api_key},
@@ -70,8 +82,11 @@ def _call_api(
                 },
             },
         ) as response:
-            for chunk in json_stream.httpx.load(response, persistent=True):
-                yield chunk["candidates"][0]["content"]["parts"][0]["text"]
+            async for chunk in ijson.items(
+                AsyncIteratorReader(response.aiter_bytes(10 * 1024)),
+                "item.candidates.item.content.parts.item.text",
+            ):
+                yield chunk
 
 
 class GeminiPro(GoogleApiAssistant):