Init

.difyignore

@@ -0,0 +1,2 @@
+../../.difyignore
+.venv

.env.example

@@ -0,0 +1,3 @@
+INSTALL_METHOD=remote
+REMOTE_INSTALL_URL=debug-plugin.dify.dev:5003
+REMOTE_INSTALL_KEY=********-****-****-****-************

.gitignore

@@ -0,0 +1,2 @@
+uv.lock
+.venv

.python-version

@@ -0,0 +1 @@
+3.12

.verification.dify.json

@@ -0,0 +1 @@
+{"authorized_category":"langgenius"}

README.md

@@ -0,0 +1,9 @@
+## Overview
+
+This plugin provides access to models that are OpenAI-compatible, including LLMs, reranking, text embedding, speech-to-text (STT), and text-to-speech(TTS) models. Developers can easily add models by providing configuration parameters such as the model name and API key.
+
+## Configure
+
+Configure the OpenAI-API-compatible model by providing its core details (Type, Name, API Key, URL) and adjusting further options like completion, context, and token limits, as well as streaming and vision settings. Save when done.
+
+![](./_assets/openai_api_compatible-01.png)

README_zh.md

@@ -0,0 +1,9 @@
+## 概述
+
+此插件提供对OpenAI兼容模型的访问，包括大语言模型(LLM)、重排序、文本嵌入、语音转文本(STT)和文本转语音(TTS)模型。开发者可以通过提供模型名称和API密钥等配置参数轻松添加模型。
+
+## 配置
+
+通过提供核心详细信息（类型、名称、API密钥、URL）来配置OpenAI-API兼容模型，并调整其他选项，如完成、上下文和令牌限制，以及流式传输和视觉设置。完成后保存。
+
+![](./_assets/openai_api_compatible-01.png)

_assets/icon.svg

@@ -0,0 +1 @@
+<svg width="16" height="17" viewBox="0 0 16 17" fill="none" xmlns="http://www.w3.org/2000/svg" class="w-4 h-4 text-gray-400" data-icon="CubeOutline" aria-hidden="true"><g id="cube-outline"><g id="Solid"><path d="M8.26865 1.29003C8.09143 1.25358 7.90866 1.25358 7.73144 1.29003C7.52659 1.33216 7.3435 1.43471 7.19794 1.51624L7.15826 1.53841L6.17628 2.08395C5.85443 2.26276 5.73846 2.66863 5.91727 2.99049C6.09608 3.31234 6.50195 3.4283 6.82381 3.24949L7.80579 2.70395C7.90681 2.64782 7.95839 2.61946 7.99686 2.60091L8.00004 2.59938L8.00323 2.60091C8.0417 2.61946 8.09327 2.64782 8.1943 2.70395L9.17628 3.24949C9.49814 3.4283 9.90401 3.31234 10.0828 2.99048C10.2616 2.66863 10.1457 2.26276 9.82381 2.08395L8.84183 1.53841L8.80215 1.51624C8.65659 1.43471 8.4735 1.33216 8.26865 1.29003Z" fill="currentColor"></path><path d="M12.8238 3.75062C12.5019 3.57181 12.0961 3.68777 11.9173 4.00963C11.7385 4.33148 11.8544 4.73735 12.1763 4.91616L12.6272 5.16668L12.1763 5.41719C11.8545 5.596 11.7385 6.00186 11.9173 6.32372C12.0961 6.64558 12.502 6.76154 12.8238 6.58273L13.3334 6.29966V6.83339C13.3334 7.20158 13.6319 7.50006 14 7.50006C14.3682 7.50006 14.6667 7.20158 14.6667 6.83339V5.79435L14.6668 5.74627C14.6673 5.62441 14.6678 5.48084 14.6452 5.33482C14.6869 5.17472 14.6696 4.99892 14.5829 4.84286C14.4904 4.6764 14.3371 4.56501 14.1662 4.52099C14.0496 4.43038 13.9239 4.36116 13.8173 4.3024L13.7752 4.27915L12.8238 3.75062Z" fill="currentColor"></path><path d="M3.8238 4.91616C4.14566 4.73735 4.26162 4.33148 4.08281 4.00963C3.90401 3.68777 3.49814 3.57181 3.17628 3.75062L2.22493 4.27915L2.18284 4.3024C2.07615 4.36116 1.95045 4.4304 1.83382 4.52102C1.66295 4.56506 1.50977 4.67643 1.41731 4.84286C1.33065 4.99886 1.31323 5.17459 1.35493 5.33464C1.33229 5.48072 1.33281 5.62436 1.33326 5.74627L1.33338 5.79435V6.83339C1.33338 7.20158 1.63185 7.50006 2.00004 7.50006C2.36823 7.50006 2.66671 7.20158 2.66671 6.83339V6.29961L3.17632 6.58273C3.49817 6.76154 3.90404 6.64558 4.08285 6.32372C4.26166 6.00186 4.1457 5.596 3.82384 5.41719L3.3729 5.16666L3.8238 4.91616Z" fill="currentColor"></path><path d="M2.66671 10.1667C2.66671 9.79853 2.36823 9.50006 2.00004 9.50006C1.63185 9.50006 1.33338 9.79853 1.33338 10.1667V11.2058L1.33326 11.2538C1.33262 11.4298 1.33181 11.6509 1.40069 11.8594C1.46024 12.0397 1.55759 12.2051 1.68622 12.3447C1.835 12.5061 2.02873 12.6128 2.18281 12.6977L2.22493 12.721L3.17628 13.2495C3.49814 13.4283 3.90401 13.3123 4.08281 12.9905C4.26162 12.6686 4.14566 12.2628 3.8238 12.084L2.87245 11.5554C2.76582 11.4962 2.71137 11.4656 2.67318 11.4413L2.66995 11.4392L2.66971 11.4354C2.66699 11.3902 2.66671 11.3277 2.66671 11.2058V10.1667Z" fill="currentColor"></path><path d="M14.6667 10.1667C14.6667 9.79853 14.3682 9.50006 14 9.50006C13.6319 9.50006 13.3334 9.79853 13.3334 10.1667V11.2058C13.3334 11.3277 13.3331 11.3902 13.3304 11.4354L13.3301 11.4392L13.3269 11.4413C13.2887 11.4656 13.2343 11.4962 13.1276 11.5554L12.1763 12.084C11.8544 12.2628 11.7385 12.6686 11.9173 12.9905C12.0961 13.3123 12.5019 13.4283 12.8238 13.2495L13.7752 12.721L13.8172 12.6977C13.9713 12.6128 14.1651 12.5061 14.3139 12.3447C14.4425 12.2051 14.5398 12.0397 14.5994 11.8594C14.6683 11.6509 14.6675 11.4298 14.6668 11.2538L14.6667 11.2058V10.1667Z" fill="currentColor"></path><path d="M6.82381 13.7506C6.50195 13.5718 6.09608 13.6878 5.91727 14.0096C5.73846 14.3315 5.85443 14.7374 6.17628 14.9162L7.15826 15.4617L7.19793 15.4839C7.29819 15.54 7.41625 15.6061 7.54696 15.6556C7.66589 15.7659 7.82512 15.8333 8.00008 15.8333C8.17507 15.8333 8.33431 15.7659 8.45324 15.6556C8.58391 15.6061 8.70193 15.54 8.80215 15.4839L8.84183 15.4617L9.82381 14.9162C10.1457 14.7374 10.2616 14.3315 10.0828 14.0096C9.90401 13.6878 9.49814 13.5718 9.17628 13.7506L8.66675 14.0337V13.5C8.66675 13.1318 8.36827 12.8333 8.00008 12.8333C7.63189 12.8333 7.33341 13.1318 7.33341 13.5V14.0337L6.82381 13.7506Z" fill="currentColor"></path><path d="M6.82384 7.08385C6.50199 6.90505 6.09612 7.02101 5.91731 7.34286C5.7385 7.66472 5.85446 8.07059 6.17632 8.2494L7.33341 8.89223V10.1666C7.33341 10.5348 7.63189 10.8333 8.00008 10.8333C8.36827 10.8333 8.66675 10.5348 8.66675 10.1666V8.89223L9.82384 8.2494C10.1457 8.07059 10.2617 7.66472 10.0829 7.34286C9.90404 7.02101 9.49817 6.90505 9.17632 7.08385L8.00008 7.73732L6.82384 7.08385Z" fill="currentColor"></path></g></g></svg>

main.py

@@ -0,0 +1,6 @@
+from dify_plugin import Plugin, DifyPluginEnv
+
+plugin = Plugin(DifyPluginEnv())
+
+if __name__ == "__main__":
+    plugin.run()

manifest.yaml

@@ -0,0 +1,31 @@
+version: 0.0.22
+type: plugin
+author: "langgenius"
+name: "openai_api_compatible"
+description:
+  en_US: Model providers compatible with OpenAI's API standard, such as LM Studio.
+  zh_Hans: 兼容 OpenAI API 的模型供应商，例如 LM Studio 。
+label:
+  en_US: "OpenAI-API-compatible"
+created_at: "2024-07-12T08:03:44.658609186Z"
+icon: icon.svg
+resource:
+  memory: 1048576
+  permission:
+    tool:
+      enabled: true
+    model:
+      enabled: true
+      llm: true
+plugins:
+  models:
+    - "provider/openai_api_compatible.yaml"
+meta:
+  version: 0.0.1
+  arch:
+    - "amd64"
+    - "arm64"
+  runner:
+    language: "python"
+    version: "3.12"
+    entrypoint: "main"

models/common_openai.py

@@ -0,0 +1,60 @@
+from collections.abc import Mapping
+
+import openai
+from httpx import Timeout
+
+from dify_plugin.errors.model import (
+    InvokeAuthorizationError,
+    InvokeBadRequestError,
+    InvokeConnectionError,
+    InvokeError,
+    InvokeRateLimitError,
+    InvokeServerUnavailableError,
+)
+
+
+class _CommonOpenAI:
+    def _to_credential_kwargs(self, credentials: Mapping) -> dict:
+        """
+        Transform credentials to kwargs for model instance
+
+        :param credentials:
+        :return:
+        """
+        credentials_kwargs = {
+            "api_key": credentials["openai_api_key"],
+            "timeout": Timeout(315.0, read=300.0, write=10.0, connect=5.0),
+            "max_retries": 1,
+        }
+
+        if credentials.get("openai_api_base"):
+            openai_api_base = credentials["openai_api_base"].rstrip("/")
+            credentials_kwargs["base_url"] = openai_api_base + "/v1"
+
+        if "openai_organization" in credentials:
+            credentials_kwargs["organization"] = credentials["openai_organization"]
+
+        return credentials_kwargs
+
+    @property
+    def _invoke_error_mapping(self) -> dict[type[InvokeError], list[type[Exception]]]:
+        """
+        Map model invoke error to unified error
+        The key is the error type thrown to the caller
+        The value is the error type thrown by the model,
+        which needs to be converted into a unified error type for the caller.
+
+        :return: Invoke error mapping
+        """
+        return {
+            InvokeConnectionError: [openai.APIConnectionError, openai.APITimeoutError],
+            InvokeServerUnavailableError: [openai.InternalServerError],
+            InvokeRateLimitError: [openai.RateLimitError],
+            InvokeAuthorizationError: [openai.AuthenticationError, openai.PermissionDeniedError],
+            InvokeBadRequestError: [
+                openai.BadRequestError,
+                openai.NotFoundError,
+                openai.UnprocessableEntityError,
+                openai.APIError,
+            ],
+        }

models/llm/llm.py

@@ -0,0 +1,167 @@
+import re
+from contextlib import suppress
+from typing import Mapping, Optional, Union, Generator
+
+from dify_plugin.entities.model import (
+    AIModelEntity,
+    DefaultParameterName,
+    I18nObject,
+    ModelFeature,
+    ParameterRule,
+    ParameterType,
+)
+from dify_plugin.entities.model.llm import LLMResult
+from dify_plugin.entities.model.message import (
+    PromptMessage,
+    PromptMessageRole,
+    PromptMessageTool,
+    SystemPromptMessage,
+    AssistantPromptMessage,
+)
+from dify_plugin.interfaces.model.openai_compatible.llm import OAICompatLargeLanguageModel
+from typing import List
+
+
+class OpenAILargeLanguageModel(OAICompatLargeLanguageModel):
+    # Pre-compiled regex for better performance
+    _THINK_PATTERN = re.compile(r"^<think>.*?</think>\s*", re.DOTALL)
+
+    def get_customizable_model_schema(
+        self, model: str, credentials: Mapping | dict
+    ) -> AIModelEntity:
+        entity = super().get_customizable_model_schema(model, credentials)
+
+        agent_though_support = credentials.get("agent_though_support", "not_supported")
+        if agent_though_support == "supported":
+            try:
+                entity.features.index(ModelFeature.AGENT_THOUGHT)
+            except ValueError:
+                entity.features.append(ModelFeature.AGENT_THOUGHT)
+
+        structured_output_support = credentials.get("structured_output_support", "not_supported")
+        if structured_output_support == "supported":
+            # ----
+            # The following section should be added after the new version of `dify-plugin-sdks`
+            # is released.
+            # Related Commit:
+            # https://github.com/langgenius/dify-plugin-sdks/commit/0690573a879caf43f92494bf411f45a1835d96f6
+            # ----
+            # try:
+            #     entity.features.index(ModelFeature.STRUCTURED_OUTPUT)
+            # except ValueError:
+            #     entity.features.append(ModelFeature.STRUCTURED_OUTPUT)
+
+            entity.parameter_rules.append(
+                ParameterRule(
+                    name=DefaultParameterName.RESPONSE_FORMAT.value,
+                    label=I18nObject(en_US="Response Format", zh_Hans="回复格式"),
+                    help=I18nObject(
+                        en_US="Specifying the format that the model must output.",
+                        zh_Hans="指定模型必须输出的格式。",
+                    ),
+                    type=ParameterType.STRING,
+                    options=["text", "json_object", "json_schema"],
+                    required=False,
+                )
+            )
+            entity.parameter_rules.append(
+                ParameterRule(
+                    name=DefaultParameterName.JSON_SCHEMA.value,
+                    use_template=DefaultParameterName.JSON_SCHEMA.value,
+                )
+            )
+
+        if "display_name" in credentials and credentials["display_name"] != "":
+            entity.label = I18nObject(
+                en_US=credentials["display_name"], zh_Hans=credentials["display_name"]
+            )
+
+        entity.parameter_rules += [
+            ParameterRule(
+                name="enable_thinking",
+                label=I18nObject(en_US="Thinking mode", zh_Hans="思考模式"),
+                help=I18nObject(
+                    en_US="Whether to enable thinking mode, applicable to various thinking mode models deployed on reasoning frameworks such as vLLM and SGLang, for example Qwen3.",
+                    zh_Hans="是否开启思考模式，适用于vLLM和SGLang等推理框架部署的多种思考模式模型，例如Qwen3。",
+                ),
+                type=ParameterType.BOOLEAN,
+                required=False,
+            )
+        ]
+        return entity
+
+    @classmethod
+    def _drop_analyze_channel(cls, prompt_messages: List[PromptMessage]) -> None:
+        """
+        Remove thinking content from assistant messages for better performance.
+
+        Uses early exit and pre-compiled regex to minimize overhead.
+        Args:
+            prompt_messages:
+
+        Returns:
+
+        """
+        for p in prompt_messages:
+            # Early exit conditions
+            if not isinstance(p, AssistantPromptMessage):
+                continue
+            if not isinstance(p.content, str):
+                continue
+            # Quick check to avoid regex if not needed
+            if not p.content.startswith("<think>"):
+                continue
+
+            # Only perform regex substitution when necessary
+            new_content = cls._THINK_PATTERN.sub("", p.content, count=1)
+            # Only update if changed
+            if new_content != p.content:
+                p.content = new_content
+
+    def _invoke(
+        self,
+        model: str,
+        credentials: dict,
+        prompt_messages: list[PromptMessage],
+        model_parameters: dict,
+        tools: Optional[list[PromptMessageTool]] = None,
+        stop: Optional[list[str]] = None,
+        stream: bool = True,
+        user: Optional[str] = None,
+    ) -> Union[LLMResult, Generator]:
+        # Compatibility adapter for Dify's 'json_schema' structured output mode.
+        # The base class does not natively handle the 'json_schema' parameter. This block
+        # translates it into a standard OpenAI-compatible request by:
+        # 1. Injecting the JSON schema directly into the system prompt to guide the model.
+        # This ensures models like gpt-4o produce the correct structured output.
+        if model_parameters.get("response_format") == "json_schema":
+            # Use .get() instead of .pop() for safety
+            json_schema_str = model_parameters.get("json_schema")
+
+            if json_schema_str:
+                structured_output_prompt = (
+                    "Your response must be a JSON object that validates against the following JSON schema, and nothing else.\n"
+                    f"JSON Schema: ```json\n{json_schema_str}\n```"
+                )
+
+                existing_system_prompt = next(
+                    (p for p in prompt_messages if p.role == PromptMessageRole.SYSTEM), None
+                )
+                if existing_system_prompt:
+                    existing_system_prompt.content = (
+                        structured_output_prompt + "\n\n" + existing_system_prompt.content
+                    )
+                else:
+                    prompt_messages.insert(0, SystemPromptMessage(content=structured_output_prompt))
+
+        enable_thinking = model_parameters.pop("enable_thinking", None)
+        if enable_thinking is not None:
+            model_parameters["chat_template_kwargs"] = {"enable_thinking": bool(enable_thinking)}
+
+        # Remove thinking content from assistant messages for better performance.
+        with suppress(Exception):
+            self._drop_analyze_channel(prompt_messages)
+
+        return super()._invoke(
+            model, credentials, prompt_messages, model_parameters, tools, stop, stream, user
+        )

models/rerank/rerank.py

@@ -0,0 +1,45 @@
+from typing import Mapping
+
+from dify_plugin.entities.model import AIModelEntity, I18nObject
+
+from dify_plugin.interfaces.model.openai_compatible.rerank import OAICompatRerankModel
+from dify_plugin.errors.model import CredentialsValidateFailedError
+
+
+class OpenAIRerankModel(OAICompatRerankModel):
+    def validate_credentials(self, model: str, credentials: dict) -> None:
+        """
+        Validate model credentials
+
+        :param model: model name
+        :param credentials: model credentials
+        :return:
+        """
+        try:
+            self._invoke(
+                model=model,
+                credentials=credentials,
+                query="What is the capital of the United States?",
+                docs=[
+                    "Carson City is the capital city of the American state of Nevada. At the 2010 United States "
+                    "Census, Carson City had a population of 55,274.",
+                    "The Commonwealth of the Northern Mariana Islands is a group of islands in the Pacific Ocean that "
+                    "are a political division controlled by the United States. Its capital is Saipan.",
+                ],
+                score_threshold=0.8,
+                top_n=3,
+            )
+        except Exception as ex:
+            raise CredentialsValidateFailedError(str(ex)) from ex
+
+    def get_customizable_model_schema(
+        self, model: str, credentials: Mapping | dict
+    ) -> AIModelEntity:
+        entity = super().get_customizable_model_schema(model, credentials)
+
+        if "display_name" in credentials and credentials["display_name"] != "":
+            entity.label = I18nObject(
+                en_US=credentials["display_name"], zh_Hans=credentials["display_name"]
+            )
+
+        return entity

models/speech2text/speech2text.py

@@ -0,0 +1,27 @@
+from typing import Optional
+from dify_plugin.entities.model import AIModelEntity, FetchFrom, I18nObject, ModelType
+from dify_plugin.interfaces.model.openai_compatible.speech2text import OAICompatSpeech2TextModel
+
+
+class OpenAISpeech2TextModel(OAICompatSpeech2TextModel):
+    def get_customizable_model_schema(
+        self, model: str, credentials: dict
+    ) -> Optional[AIModelEntity]:
+        """
+        used to define customizable model schema
+        """
+        entity = AIModelEntity(
+            model=model,
+            label=I18nObject(en_US=model),
+            fetch_from=FetchFrom.CUSTOMIZABLE_MODEL,
+            model_type=ModelType.SPEECH2TEXT,
+            model_properties={},
+            parameter_rules=[],
+        )
+
+        if "display_name" in credentials and credentials["display_name"] != "":
+            entity.label = I18nObject(
+                en_US=credentials["display_name"], zh_Hans=credentials["display_name"]
+            )
+
+        return entity

models/text_embedding/text_embedding.py

@@ -0,0 +1,21 @@
+from typing import Mapping
+
+from dify_plugin.entities.model import AIModelEntity, I18nObject
+
+from dify_plugin.interfaces.model.openai_compatible.text_embedding import OAICompatEmbeddingModel
+
+
+class OpenAITextEmbeddingModel(OAICompatEmbeddingModel):
+
+    def get_customizable_model_schema(
+        self, model: str, credentials: Mapping | dict
+    ) -> AIModelEntity:
+        credentials = credentials or {}
+        entity = super().get_customizable_model_schema(model, credentials)
+
+        if "display_name" in credentials and credentials["display_name"] != "":
+            entity.label = I18nObject(
+                en_US=credentials["display_name"], zh_Hans=credentials["display_name"]
+            )
+
+        return entity

models/tts/tts.py

@@ -0,0 +1,20 @@
+from typing import Mapping
+
+from dify_plugin.entities.model import AIModelEntity, I18nObject
+
+from dify_plugin.interfaces.model.openai_compatible.tts import OAICompatText2SpeechModel
+
+
+class OpenAIText2SpeechModel(OAICompatText2SpeechModel):
+
+    def get_customizable_model_schema(
+        self, model: str, credentials: Mapping | dict
+    ) -> AIModelEntity:
+        entity = super().get_customizable_model_schema(model, credentials)
+
+        if "display_name" in credentials and credentials["display_name"] != "":
+            entity.label = I18nObject(
+                en_US=credentials["display_name"], zh_Hans=credentials["display_name"]
+            )
+
+        return entity

provider/openai_api_compatible.py

@@ -0,0 +1,17 @@
+import logging
+from collections.abc import Mapping
+
+from dify_plugin import ModelProvider
+
+logger = logging.getLogger(__name__)
+
+
+class OpenAIProvider(ModelProvider):
+    def validate_provider_credentials(self, credentials: Mapping) -> None:
+        """
+        Validate provider credentials
+        if validate failed, raise exception
+
+        :param credentials: provider credentials, credentials form defined in `provider_credential_schema`.
+        """
+        pass

provider/openai_api_compatible.yaml

@@ -0,0 +1,276 @@
+provider: openai_api_compatible
+label:
+  en_US: OpenAI-API-compatible
+description:
+  en_US: Model providers compatible with OpenAI's API standard, such as LM Studio.
+  zh_Hans: 兼容 OpenAI API 的模型供应商，例如 LM Studio 。
+icon_small:
+  en_US: icon.svg
+supported_model_types:
+  - llm
+  - rerank
+  - text-embedding
+  - speech2text
+  - tts
+configurate_methods:
+  - customizable-model
+model_credential_schema:
+  model:
+    label:
+      en_US: Model Name
+      zh_Hans: 模型名称
+    placeholder:
+      en_US: Enter full model name
+      zh_Hans: 输入模型全称
+  credential_form_schemas:
+    - variable: display_name
+      label:
+        en_US: Model display name
+        zh_Hans: 模型显示名称
+      type: text-input
+      required: false
+      placeholder:
+        zh_Hans: 模型在界面的显示名称
+        en_US: The display name of the model in the interface.
+    - variable: api_key
+      label:
+        en_US: API Key
+      type: secret-input
+      required: false
+      placeholder:
+        zh_Hans: 在此输入您的 API Key
+        en_US: Enter your API Key
+    - variable: endpoint_url
+      label:
+        zh_Hans: API endpoint URL
+        en_US: API endpoint URL
+      type: text-input
+      required: true
+      placeholder:
+        zh_Hans: Base URL, e.g. https://api.openai.com/v1
+        en_US: Base URL, e.g. https://api.openai.com/v1
+    - variable: endpoint_model_name
+      label:
+        zh_Hans: API endpoint中的模型名称
+        en_US: model name for API endpoint
+      type: text-input
+      required: false
+      placeholder:
+        zh_Hans: endpoint model name, e.g. chatgpt4.0
+        en_US: endpoint model name, e.g. chatgpt4.0
+    - variable: mode
+      show_on:
+        - variable: __model_type
+          value: llm
+      label:
+        en_US: Completion mode
+      type: select
+      required: false
+      default: chat
+      placeholder:
+        zh_Hans: 选择对话类型
+        en_US: Select completion mode
+      options:
+        - value: completion
+          label:
+            en_US: Completion
+            zh_Hans: 补全
+        - value: chat
+          label:
+            en_US: Chat
+            zh_Hans: 对话
+    - variable: context_size
+      label:
+        zh_Hans: 模型上下文长度
+        en_US: Model context size
+      required: true
+      show_on:
+        - variable: __model_type
+          value: llm
+      type: text-input
+      default: "4096"
+      placeholder:
+        zh_Hans: 在此输入您的模型上下文长度
+        en_US: Enter your Model context size
+    - variable: context_size
+      label:
+        zh_Hans: 模型上下文长度
+        en_US: Model context size
+      required: true
+      show_on:
+        - variable: __model_type
+          value: text-embedding
+      type: text-input
+      default: "4096"
+      placeholder:
+        zh_Hans: 在此输入您的模型上下文长度
+        en_US: Enter your Model context size
+    - variable: context_size
+      label:
+        zh_Hans: 模型上下文长度
+        en_US: Model context size
+      required: true
+      show_on:
+        - variable: __model_type
+          value: rerank
+      type: text-input
+      default: "4096"
+      placeholder:
+        zh_Hans: 在此输入您的模型上下文长度
+        en_US: Enter your Model context size
+    - variable: max_tokens_to_sample
+      label:
+        zh_Hans: 最大 token 上限
+        en_US: Upper bound for max tokens
+      show_on:
+        - variable: __model_type
+          value: llm
+      default: "4096"
+      type: text-input
+    - variable: agent_though_support
+      show_on:
+        - variable: __model_type
+          value: llm
+      label:
+        en_US: Agent Thought
+      type: select
+      required: false
+      default: not_supported
+      options:
+        - value: supported
+          label:
+            en_US: Support
+            zh_Hans: 支持
+        - value: not_supported
+          label:
+            en_US: Not Support
+            zh_Hans: 不支持
+    - variable: function_calling_type
+      show_on:
+        - variable: __model_type
+          value: llm
+      label:
+        en_US: Function calling
+      type: select
+      required: false
+      default: no_call
+      options:
+        - value: function_call
+          label:
+            en_US: Function Call
+            zh_Hans: Function Call
+        - value: tool_call
+          label:
+            en_US: Tool Call
+            zh_Hans: Tool Call
+        - value: no_call
+          label:
+            en_US: Not Support
+            zh_Hans: 不支持
+    - variable: stream_function_calling
+      show_on:
+        - variable: __model_type
+          value: llm
+      label:
+        en_US: Stream function calling
+      type: select
+      required: false
+      default: not_supported
+      options:
+        - value: supported
+          label:
+            en_US: Support
+            zh_Hans: 支持
+        - value: not_supported
+          label:
+            en_US: Not Support
+            zh_Hans: 不支持
+    - variable: vision_support
+      show_on:
+        - variable: __model_type
+          value: llm
+      label:
+        zh_Hans: Vision 支持
+        en_US: Vision Support
+      type: select
+      required: false
+      default: no_support
+      options:
+        - value: support
+          label:
+            en_US: Support
+            zh_Hans: 支持
+        - value: no_support
+          label:
+            en_US: Not Support
+            zh_Hans: 不支持
+    - variable: structured_output_support
+      show_on:
+        - variable: __model_type
+          value: llm
+      label:
+        en_US: Structured Output
+      type: select
+      required: false
+      default: not_supported
+      options:
+        - value: supported
+          label:
+            en_US: Support
+            zh_Hans: 支持
+        - value: not_supported
+          label:
+            en_US: Not Support
+            zh_Hans: 不支持
+    - variable: stream_mode_auth
+      show_on:
+        - variable: __model_type
+          value: llm
+      label:
+        en_US: Stream mode auth
+      type: select
+      required: false
+      default: not_use
+      options:
+        - value: use
+          label:
+            en_US: Use
+            zh_Hans: 使用
+        - value: not_use
+          label:
+            en_US: Not Use
+            zh_Hans: 不使用
+    - variable: stream_mode_delimiter
+      label:
+        zh_Hans: 流模式返回结果的分隔符
+        en_US: Delimiter for streaming results
+      show_on:
+        - variable: __model_type
+          value: llm
+      default: '\n\n'
+      type: text-input
+    - variable: voices
+      show_on:
+        - variable: __model_type
+          value: tts
+      label:
+        en_US: Available Voices (comma-separated)
+        zh_Hans: 可用声音（用英文逗号分隔）
+      type: text-input
+      required: false
+      default: "alloy"
+      placeholder:
+        en_US: "alloy,echo,fable,onyx,nova,shimmer"
+        zh_Hans: "alloy,echo,fable,onyx,nova,shimmer"
+      help:
+        en_US: "List voice names separated by commas. First voice will be used as default."
+        zh_Hans: "用英文逗号分隔的声音列表。第一个声音将作为默认值。"
+extra:
+  python:
+    provider_source: provider/openai_api_compatible.py
+    model_sources:
+      - "models/llm/llm.py"
+      - "models/text_embedding/text_embedding.py"
+      - "models/rerank/rerank.py"
+      - "models/speech2text/speech2text.py"
+      - "models/tts/tts.py"

pyproject.toml

@@ -0,0 +1,21 @@
+[project]
+name = "openai-api-compatible-GnRzR6cEy"
+version = "0.1.0"
+description = "Add your description here"
+readme = "README.md"
+requires-python = ">=3.12"
+
+# uv pip compile pyproject.toml -o ./requirements.txt
+dependencies = [
+    "dify-plugin>=0.4.3",
+    "openai>=1.99.9",
+]
+
+# uv run black . -C -l 100
+# uv run ruff check --fix
+[dependency-groups]
+dev = [
+    "black>=25.1.0",
+    "pytest>=8.4.1",
+    "ruff>=0.12.5",
+]

requirements.txt

@@ -0,0 +1,127 @@
+# This file was autogenerated by uv via the following command:
+#    uv pip compile pyproject.toml -o ./requirements.txt
+annotated-types==0.7.0
+    # via pydantic
+anyio==4.10.0
+    # via
+    #   httpx
+    #   openai
+blinker==1.9.0
+    # via flask
+certifi==2025.8.3
+    # via
+    #   httpcore
+    #   httpx
+    #   requests
+cffi==1.17.1
+    # via gevent
+charset-normalizer==3.4.3
+    # via requests
+click==8.2.1
+    # via flask
+colorama==0.4.6
+    # via
+    #   click
+    #   tqdm
+dify-plugin==0.4.3
+    # via openai-api-compatible-gnrzr6cey (pyproject.toml)
+distro==1.9.0
+    # via openai
+dpkt==1.9.8
+    # via dify-plugin
+flask==3.0.3
+    # via dify-plugin
+gevent==25.5.1
+    # via dify-plugin
+greenlet==3.2.4
+    # via gevent
+h11==0.16.0
+    # via httpcore
+httpcore==1.0.9
+    # via httpx
+httpx==0.28.1
+    # via
+    #   dify-plugin
+    #   openai
+idna==3.10
+    # via
+    #   anyio
+    #   httpx
+    #   requests
+    #   yarl
+itsdangerous==2.2.0
+    # via flask
+jinja2==3.1.6
+    # via flask
+jiter==0.10.0
+    # via openai
+markupsafe==3.0.2
+    # via
+    #   jinja2
+    #   werkzeug
+multidict==6.6.4
+    # via yarl
+openai==1.99.9
+    # via openai-api-compatible-gnrzr6cey (pyproject.toml)
+packaging==25.0
+    # via dify-plugin
+propcache==0.3.2
+    # via yarl
+pycparser==2.22
+    # via cffi
+pydantic==2.11.7
+    # via
+    #   dify-plugin
+    #   openai
+    #   pydantic-settings
+pydantic-core==2.33.2
+    # via pydantic
+pydantic-settings==2.10.1
+    # via dify-plugin
+python-dotenv==1.1.1
+    # via pydantic-settings
+pyyaml==6.0.2
+    # via dify-plugin
+regex==2025.7.34
+    # via tiktoken
+requests==2.32.4
+    # via
+    #   dify-plugin
+    #   tiktoken
+setuptools==80.9.0
+    # via
+    #   zope-event
+    #   zope-interface
+sniffio==1.3.1
+    # via
+    #   anyio
+    #   openai
+socksio==1.0.0
+    # via dify-plugin
+tiktoken==0.8.0
+    # via dify-plugin
+tqdm==4.67.1
+    # via openai
+typing-extensions==4.14.1
+    # via
+    #   anyio
+    #   openai
+    #   pydantic
+    #   pydantic-core
+    #   typing-inspection
+typing-inspection==0.4.1
+    # via
+    #   pydantic
+    #   pydantic-settings
+urllib3==2.5.0
+    # via requests
+werkzeug==3.0.6
+    # via
+    #   dify-plugin
+    #   flask
+yarl==1.20.1
+    # via dify-plugin
+zope-event==5.1.1
+    # via gevent
+zope-interface==7.2
+    # via gevent