A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 8, 15:15 UTC
[CI/Build] Relax `test_mcp_tool_call` (#34204)
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
Cyrus Leung committed
Feb 10, 2026 at 13:18 UTC
ab97bcf66295fca10a892bd14090e902b4b3c317
1 file changed
+3
-3
tests/entrypoints/openai/responses/test_parsable_context.py
+3
-3
@@ -1,7 +1,7 @@
1
# SPDX-License-Identifier: Apache-2.0
2
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
3
4
-import importlib
4
+import importlib.util
5
import json
6
7
import pytest
@@ -179,12 +179,12 @@ async def test_mcp_tool_call(client: OpenAI, model_name: str):
179
assert response.output[2].type == "reasoning"
180
# make sure the correct math is in the final output
181
assert response.output[3].type == "message"
182
- assert "56088" in response.output[3].content[0].text
182
+ assert any(s in response.output[3].content[0].text for s in ("56088", "56,088"))
183
184
# test raw input_messages / output_messages
185
assert len(response.input_messages) == 1
186
assert len(response.output_messages) == 3
187
- assert "56088" in response.output_messages[2]["message"]
187
+ assert any(s in response.output_messages[2]["message"] for s in ("56088", "56,088"))
188
189
190
@pytest.mark.asyncio