diff --git a/sentry_sdk/integrations/pydantic_ai/spans/invoke_agent.py b/sentry_sdk/integrations/pydantic_ai/spans/invoke_agent.py
index 03cb707141..e9fdc90fe0 100644
--- a/sentry_sdk/integrations/pydantic_ai/spans/invoke_agent.py
+++ b/sentry_sdk/integrations/pydantic_ai/spans/invoke_agent.py
@@ -19,7 +19,6 @@
 from .utils import (
     _serialize_binary_content_item,
     _serialize_image_url_item,
-    _set_usage_data,
 )
 
 if TYPE_CHECKING:
@@ -150,16 +149,6 @@ def update_invoke_agent_span(span: "sentry_sdk.tracing.Span", result: "Any") ->
             span, SPANDATA.GEN_AI_RESPONSE_TEXT, str(output), unpack=False
         )
 
-    # Set token usage data if available
-    if hasattr(result, "usage") and callable(result.usage):
-        try:
-            usage = result.usage()
-            if usage:
-                _set_usage_data(span, usage)
-        except Exception:
-            # If usage() call fails, continue without setting usage data
-            pass
-
     # Set model name from response if available
     if hasattr(result, "response"):
         try:
diff --git a/tests/integrations/pydantic_ai/test_pydantic_ai.py b/tests/integrations/pydantic_ai/test_pydantic_ai.py
index 9a1fe79167..7aad94d852 100644
--- a/tests/integrations/pydantic_ai/test_pydantic_ai.py
+++ b/tests/integrations/pydantic_ai/test_pydantic_ai.py
@@ -205,71 +205,6 @@ def failing_model(messages, info):
         assert spans[0]["status"] == "internal_error"
 
 
-@pytest.mark.parametrize("stream_gen_ai_spans", [True, False])
-@pytest.mark.asyncio
-async def test_agent_run_async_usage_data(
-    sentry_init,
-    capture_events,
-    capture_items,
-    get_test_agent,
-    stream_gen_ai_spans,
-):
-    """
-    Test that the invoke_agent span includes token usage and model data.
-    """
-    sentry_init(
-        integrations=[PydanticAIIntegration()],
-        traces_sample_rate=1.0,
-        send_default_pii=True,
-        stream_gen_ai_spans=stream_gen_ai_spans,
-    )
-
-    test_agent = get_test_agent()
-
-    if stream_gen_ai_spans:
-        items = capture_items("transaction", "span")
-
-        result = await test_agent.run("Test input")
-
-        assert result is not None
-        assert result.output is not None
-
-        (transaction,) = (item.payload for item in items if item.type == "transaction")
-    else:
-        events = capture_events()
-
-        result = await test_agent.run("Test input")
-
-        assert result is not None
-        assert result.output is not None
-
-        (transaction,) = events
-
-    # Verify transaction (the transaction IS the invoke_agent span)
-    assert transaction["transaction"] == "invoke_agent test_agent"
-
-    # The invoke_agent span should have token usage data
-    trace_data = transaction["contexts"]["trace"].get("data", {})
-    assert "gen_ai.usage.input_tokens" in trace_data, (
-        "Missing input_tokens on invoke_agent span"
-    )
-    assert "gen_ai.usage.output_tokens" in trace_data, (
-        "Missing output_tokens on invoke_agent span"
-    )
-    assert "gen_ai.usage.total_tokens" in trace_data, (
-        "Missing total_tokens on invoke_agent span"
-    )
-    assert "gen_ai.response.model" in trace_data, (
-        "Missing response.model on invoke_agent span"
-    )
-
-    # Verify the values are reasonable
-    assert trace_data["gen_ai.usage.input_tokens"] > 0
-    assert trace_data["gen_ai.usage.output_tokens"] > 0
-    assert trace_data["gen_ai.usage.total_tokens"] > 0
-    assert trace_data["gen_ai.response.model"] == "test"  # Test model name
-
-
 @pytest.mark.parametrize("stream_gen_ai_spans", [True, False])
 def test_agent_run_sync(
     sentry_init,