From 01ec67a98cddbd759ca8a9e95657d89aaf72d736 Mon Sep 17 00:00:00 2001
From: Yeuoly <45712896+Yeuoly@users.noreply.github.com>
Date: Mon, 22 Dec 2025 19:03:02 +0800
Subject: [PATCH 1/2] Bump SDK version to 0.7.1

---
 .../interfaces/model/openai_compatible/llm.py         | 11 +++++++++--
 python/pdm.lock                                       |  2 +-
 python/pyproject.toml                                 |  2 +-
 3 files changed, 11 insertions(+), 4 deletions(-)

diff --git a/python/dify_plugin/interfaces/model/openai_compatible/llm.py b/python/dify_plugin/interfaces/model/openai_compatible/llm.py
index dcff5ffd..92c02924 100644
--- a/python/dify_plugin/interfaces/model/openai_compatible/llm.py
+++ b/python/dify_plugin/interfaces/model/openai_compatible/llm.py
@@ -183,7 +183,11 @@ def validate_credentials(self, model: str, credentials: dict) -> None:
                 endpoint_url += "/"
 
             # prepare the payload for a simple ping to the model
-            data = {"model": credentials.get("endpoint_model_name", model), "max_tokens": 5}
+            validate_credentials_max_tokens = credentials.get("validate_credentials_max_tokens", 5) or 5
+            data = {
+                "model": credentials.get("endpoint_model_name", model),
+                "max_tokens": validate_credentials_max_tokens,
+            }
 
             completion_type = LLMMode.value_of(credentials["mode"])
 
@@ -201,8 +205,11 @@ def validate_credentials(self, model: str, credentials: dict) -> None:
             # ADD stream validate_credentials
             stream_mode_auth = credentials.get("stream_mode_auth", "not_use")
             if stream_mode_auth == "use":
+                stream_validate_max_tokens = credentials.get("validate_credentials_max_tokens", None) or 10
                 data["stream"] = True
-                data["max_tokens"] = 10
+                # default 10 (introduced in PR #93) to ensure streaming endpoints emit a token chunk;
+                # allow overriding via credentials when explicitly provided.
+                data["max_tokens"] = stream_validate_max_tokens
                 response = requests.post(endpoint_url, headers=headers, json=data, timeout=(10, 300), stream=True)
                 if response.status_code != 200:
                     raise CredentialsValidateFailedError(
diff --git a/python/pdm.lock b/python/pdm.lock
index 006cbbe2..a4d8df11 100644
--- a/python/pdm.lock
+++ b/python/pdm.lock
@@ -12,7 +12,7 @@ requires_python = ">=3.11"
 
 [[package]]
 name = "annotated-types"
-version = "0.7.0"
+version = "0.7.1"
 requires_python = ">=3.8"
 summary = "Reusable constraint types to use with typing.Annotated"
 groups = ["default"]
diff --git a/python/pyproject.toml b/python/pyproject.toml
index c39c85e0..958e2e2d 100644
--- a/python/pyproject.toml
+++ b/python/pyproject.toml
@@ -1,6 +1,6 @@
 [project]
 name = "dify_plugin"
-version = "0.7.0"
+version = "0.7.1"
 description = "Dify Plugin SDK"
 authors = [{ name = "langgenius", email = "hello@dify.ai" }]
 dependencies = [

From a6d965f41a6bf714154037b3cc869bf486a31b58 Mon Sep 17 00:00:00 2001
From: Yeuoly <45712896+Yeuoly@users.noreply.github.com>
Date: Thu, 25 Dec 2025 00:12:55 +0800
Subject: [PATCH 2/2] Revert annotated-types lock version change

---
 python/pdm.lock | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/python/pdm.lock b/python/pdm.lock
index a4d8df11..006cbbe2 100644
--- a/python/pdm.lock
+++ b/python/pdm.lock
@@ -12,7 +12,7 @@ requires_python = ">=3.11"
 
 [[package]]
 name = "annotated-types"
-version = "0.7.1"
+version = "0.7.0"
 requires_python = ">=3.8"
 summary = "Reusable constraint types to use with typing.Annotated"
 groups = ["default"]