MeUtils 2025.8.29.20.5.48__py3-none-any.whl → 2025.9.3.23.13.47__py3-none-any.whl

meutils/llm/openai_polling/chat.py

@@ -34,18 +34,61 @@ class Completions(object):
         self.client = AsyncOpenAI(base_url=self.base_url, api_key=self.api_key, http_client=http_client)
 
     async def create(self, request: CompletionRequest):
+        # todo:
+        # 1. 可思考模型 思考优先
+        # request.thinking = {"type": "enabled"}
+
+        if request.thinking and request.model.startswith("doubao-seed") and request.thinking.get("type") == "disabled":
+            choices = {
+                          "doubao-seed-1.6-250615",
+                          "doubao-seed-1-6-250615",
+                          "doubao-seed-1-6-vision-250815"
+                      } - {request.model}
+            request.model = np.random.choice(list(choices))
+        elif request.model.startswith("doubao-seed"):
+            request.thinking = {"type": "enabled"}
+            choices = {
+                          "doubao-seed-1.6-250615",
+                          "doubao-seed-1-6-250615",
+                          "doubao-seed-1-6-vision-250815",
+                          "doubao-seed-1-6-thinking-250615",
+                          "doubao-seed-1-6-thinking-250715"
+                      } - {request.model}
+            request.model = np.random.choice(list(choices))
+
+
+        elif request.model.startswith(("doubao-1-5-thinking",)):  # thinking
+            request.thinking = None
+
+            choices = {
+                          "doubao-1-5-thinking-pro-250415",
+                          "doubao-1-5-thinking-pro-m-250428",
+                          "doubao-1-5-thinking-vision-pro-250428",
+                          "doubao-1-5-ui-tars-250428",
+                      } - {request.model}
+            request.model = np.random.choice(list(choices))
+
+        elif request.model.startswith(("doubao-1-5", "doubao-1.5")):  # nothinking
+            request.thinking = {"type": "disabled"}
+
+            choices = {
+                          "doubao-1-5-pro-32k-250115",
+                          "doubao-1-5-pro-256k-250115",
+                          "doubao-1-5-pro-32k-character-250715",
+                          "doubao-1-5-pro-32k-character-250228",
+                          "doubao-1.5-vision-pro-250328",
+                          "doubao-1-5-vision-pro-32k-250115",
+                          "doubao-1-5-thinking-pro-250415",
+                          "doubao-1-5-thinking-pro-m-250428",
+                          "doubao-1-5-thinking-vision-pro-250428",
+                          "doubao-1-5-ui-tars-250428",
+                      } - {request.model}
+            request.model = np.random.choice(list(choices))
+        # todo vision
 
-        if request.model.startswith("doubao-seed"):  # todo: 1-5
-            choices = """
-            doubao-seed-1-6-250615
-            doubao-seed-1-6-thinking-250615
-            doubao-seed-1-6-thinking-250715
-            doubao-seed-1-6-vision-250815
-            """.split()
-            request.model = np.random.choice(choices)
         ###########################################################################
         # 开启视觉模型
-        if not any(i in request.model.lower() for i in ["v-", "vl", 'vision', 'doubao-seed']) and (
+        if not any(i in request.model.lower() for i in ["v-", "vl", "ui", 'vision', 'doubao-seed']) and (
                 urls := request.last_urls.get("image_url")):
             # logger.debug(request)
             if request.model.startswith(("gemini",)):
@@ -125,8 +168,12 @@ if __name__ == '__main__':
         # model="gemini-2.0-flash",
         # model="glm-4-flash",
         # model="deepseek-ai/DeepSeek-V3",
-        model="Qwen/Qwen3-8B1",
-
+        # model="Qwen/Qwen3-8B1",
+        # model="Qwen/Qwen3-8B1",
+        # model="doubao-1-5-thinking",
+        # model="doubao-1-5",
+        model="doubao-seed",
+        thinking={"type": "disabled"},
         messages=[
             {
                 "role": "system",
@@ -141,7 +188,7 @@ if __name__ == '__main__':
             {"role": "user", "content": [
                 {
                     "type": "text",
-                    "text": "解释下" * 30000
+                    "text": "解释下" * 1
                 },
                 # {
                 #     "image_url": {
@@ -153,7 +200,7 @@ if __name__ == '__main__':
             ]}
         ],
         stream=False,
-        max_tokens=None,
+        max_tokens=10,
     )
     # arun(Completions().create(request))
     # d = {
@@ -222,6 +269,10 @@ if __name__ == '__main__':
     #     stream=False,
     #     max_tokens=None,
     # )
-    api_key = os.getenv("SILICONFLOW_API_KEY")
-    base_url = "https://api.siliconflow.cn"
+    # api_key = os.getenv("SILICONFLOW_API_KEY")
+    # base_url = "https://api.siliconflow.cn"
+    # arun(Completions(api_key=api_key, base_url=base_url).create(request))
+
+    api_key = os.getenv("VOLC_API_KEY")
+    base_url = os.getenv("VOLC_BASE_URL")
     arun(Completions(api_key=api_key, base_url=base_url).create(request))

meutils/schemas/oneapi/common.py

@@ -101,6 +101,11 @@ FAL_MODELS = {
     "fal-ai/wan/v2.2-5b/text-to-video": 0.15 * FAL_,
     "fal-ai/wan/v2.2-5b/image-to-video": 0.15 * FAL_,
 
+    # pika
+    "fal-ai/pika/v2.2/text-to-video": 0.45 * FAL,
+    "fal-ai/pika/v2.2/image-to-video": 0.45 * FAL,
+    "fal-ai/pika/v2.2/pikascenes": 0.45 * FAL,
+
 }
 
 FAL_MODELS = {

meutils/serving/fastapi/gunicorn.conf.py

@@ -78,10 +78,10 @@ backlog = 1024
 
 workers = int(os.getenv("FASTAPI_WORKERS", 2))
 worker_class = "uvicorn.workers.UvicornWorker"
-worker_connections = 1000
-timeout = 300
-graceful_timeout = 100  # 在接收到重新启动信号后，worker有这么多时间来完成服务请求。超时后（从接收到重启信号开始）仍然活着的worder被强制杀死。
-keepalive = 5
+worker_connections = 1024
+timeout = 360
+graceful_timeout = 128  # 在接收到重新启动信号后，worker有这么多时间来完成服务请求。超时后（从接收到重启信号开始）仍然活着的worder被强制杀死。
+keepalive = 16
 reload = False
 
 #