diff --git a/models/openai/models/common_openai.py b/models/openai/models/common_openai.py
index cc0ff00e5..26c114888 100644
--- a/models/openai/models/common_openai.py
+++ b/models/openai/models/common_openai.py
@@ -1,5 +1,6 @@
 from collections.abc import Generator, Iterable, Mapping
 import hashlib
+import re
 from typing import TypeVar
 
 import openai
@@ -19,10 +20,42 @@ def _user_digest(user: str) -> str:
     return hashlib.sha256(user.encode()).hexdigest()
 
 
+def _chat_completions_tools_reasoning_guidance(error: Exception) -> str | None:
+    if not isinstance(error, openai.BadRequestError):
+        return None
+
+    body = error.body
+    if isinstance(body, Mapping):
+        nested_error = body.get("error")
+        if isinstance(nested_error, Mapping):
+            body = nested_error
+        body_message = body.get("message")
+    else:
+        body_message = None
+
+    message = body_message if isinstance(body_message, str) else str(error)
+    normalized_message = message.lower()
+    if not all(
+        token in normalized_message
+        for token in ("function tools", "reasoning_effort", "/v1/chat/completions")
+    ):
+        return None
+
+    model_match = re.search(r"\bgpt-(\d+(?:\.\d+)*)", message, re.IGNORECASE)
+    model_label = f"GPT-{model_match.group(1)}" if model_match else "The selected model"
+    return (
+        f"{model_label} cannot use tools with reasoning enabled through the Chat Completions API.\n"
+        "To fix this, open the model's API Key Authorization Configuration, change API Protocol "
+        "from Chat Completions to Responses, save the configuration, and try again."
+    )
+
+
 class _CommonOpenAI:
     def _transform_invoke_error(self, error: Exception) -> InvokeError:
         if isinstance(error, InvokeError):
             return error
+        if guidance := _chat_completions_tools_reasoning_guidance(error):
+            return InvokeBadRequestError(guidance)
         return super()._transform_invoke_error(error)
 
     def _to_credential_kwargs(self, credentials: Mapping) -> dict:
diff --git a/models/openai/tests/test_routing.py b/models/openai/tests/test_routing.py
index af6c02734..5d165582c 100644
--- a/models/openai/tests/test_routing.py
+++ b/models/openai/tests/test_routing.py
@@ -2,6 +2,8 @@
 
 import hashlib
 
+import httpx
+import openai
 import pytest
 
 from dify_plugin.entities.model.message import PromptMessageTool
@@ -10,6 +12,29 @@
 from models.llm.llm import _base_model, _uses_responses
 
 
+_CHAT_TOOLS_REASONING_ERROR = (
+    "Function tools with reasoning_effort are not supported for gpt-5.6-sol in "
+    "/v1/chat/completions. To use function tools, use /v1/responses or set "
+    "reasoning_effort to 'none'."
+)
+_CHAT_TOOLS_REASONING_GUIDANCE = (
+    "GPT-5.6 cannot use tools with reasoning enabled through the Chat Completions API.\n"
+    "To fix this, open the model's API Key Authorization Configuration, change API Protocol "
+    "from Chat Completions to Responses, save the configuration, and try again."
+)
+
+
+def _bad_request_error(body: object) -> openai.BadRequestError:
+    return openai.BadRequestError(
+        "Error code: 400",
+        response=httpx.Response(
+            400,
+            request=httpx.Request("POST", "https://api.openai.com/v1/chat/completions"),
+        ),
+        body=body,
+    )
+
+
 def test_reasoning_parameters_are_merged_without_mutating_the_caller() -> None:
     source = {
         "reasoning": {"effort": "low"},
@@ -310,3 +335,43 @@ def broken_stream():
 
     assert caught.value is transformed
     transform.assert_called_once_with(source)
+
+
+@pytest.mark.parametrize(
+    "body",
+    [
+        {
+            "message": _CHAT_TOOLS_REASONING_ERROR,
+            "type": "invalid_request_error",
+            "param": "reasoning_effort",
+        },
+        {
+            "error": {
+                "message": _CHAT_TOOLS_REASONING_ERROR,
+                "type": "invalid_request_error",
+                "param": "reasoning_effort",
+            }
+        },
+    ],
+)
+def test_chat_tools_reasoning_error_has_actionable_protocol_guidance(llm, body) -> None:
+    transformed = llm._transform_invoke_error(_bad_request_error(body))
+
+    assert isinstance(transformed, InvokeBadRequestError)
+    assert str(transformed) == _CHAT_TOOLS_REASONING_GUIDANCE
+
+
+def test_unrelated_bad_request_keeps_standard_error_mapping(llm) -> None:
+    transformed = llm._transform_invoke_error(
+        _bad_request_error(
+            {
+                "message": "Unsupported parameter: temperature",
+                "type": "invalid_request_error",
+                "param": "temperature",
+            }
+        )
+    )
+
+    assert isinstance(transformed, InvokeBadRequestError)
+    assert "Bad Request Error" in str(transformed)
+    assert "API Key Authorization Configuration" not in str(transformed)