Merge pull request browser-use#294 from fyq163/my_dev

Adding Moonshot model and correct image extension

Author: warmshao

.env.example

@@ -21,6 +21,9 @@ OLLAMA_ENDPOINT=http://localhost:11434
 ALIBABA_ENDPOINT=https://dashscope.aliyuncs.com/compatible-mode/v1
 ALIBABA_API_KEY=
 
+MOONSHOT_ENDPOINT=https://api.moonshot.cn/v1
+MOONSHOT_API_KEY=
+
 # Set to false to disable anonymized telemetry
 ANONYMIZED_TELEMETRY=true
 
@@ -44,4 +47,4 @@ RESOLUTION_WIDTH=1920
 RESOLUTION_HEIGHT=1080
 
 # VNC settings
-VNC_PASSWORD=youvncpassword
+VNC_PASSWORD=youvncpassword

src/utils/utils.py

@@ -20,7 +20,8 @@
     "anthropic": "Anthropic",
     "deepseek": "DeepSeek",
     "google": "Google",
-    "alibaba": "Alibaba"
+    "alibaba": "Alibaba",
+    "moonshot": "MoonShot"
 }
 
 def get_llm_model(provider: str, **kwargs):
@@ -149,6 +150,14 @@ def get_llm_model(provider: str, **kwargs):
             base_url=base_url,
             api_key=api_key,
         )
+
+    elif provider == "moonshot":
+        return ChatOpenAI(
+            model=kwargs.get("model_name", "moonshot-v1-32k-vision-preview"),
+            temperature=kwargs.get("temperature", 0.0),
+            base_url=os.getenv("MOONSHOT_ENDPOINT"),
+            api_key=os.getenv("MOONSHOT_API_KEY"),
+        )
     else:
         raise ValueError(f"Unsupported provider: {provider}")
 

tests/test_llm_api.py

@@ -22,15 +22,14 @@ class LLMConfig:
 
 def create_message_content(text, image_path=None):
     content = [{"type": "text", "text": text}]
-
+    image_format = "png" if image_path and image_path.endswith(".png") else "jpeg"
     if image_path:
         from src.utils import utils
         image_data = utils.encode_image(image_path)
         content.append({
             "type": "image_url",
-            "image_url": {"url": f"data:image/jpeg;base64,{image_data}"}
+            "image_url": {"url": f"data:image/{image_format};base64,{image_data}"}
         })
-
     return content
 
 def get_env_value(key, provider):
@@ -41,6 +40,7 @@ def get_env_value(key, provider):
         "deepseek": {"api_key": "DEEPSEEK_API_KEY", "base_url": "DEEPSEEK_ENDPOINT"},
         "mistral": {"api_key": "MISTRAL_API_KEY", "base_url": "MISTRAL_ENDPOINT"},
         "alibaba": {"api_key": "ALIBABA_API_KEY", "base_url": "ALIBABA_ENDPOINT"},
+        "moonshot":{"api_key": "MOONSHOT_API_KEY", "base_url": "MOONSHOT_ENDPOINT"},
     }
 
     if provider in env_mappings and key in env_mappings[provider]:
@@ -122,6 +122,10 @@ def test_mistral_model():
     config = LLMConfig(provider="mistral", model_name="pixtral-large-latest")
     test_llm(config, "Describe this image", "assets/examples/test.png")
 
+def test_moonshot_model():
+    config = LLMConfig(provider="moonshot", model_name="moonshot-v1-32k-vision-preview")
+    test_llm(config, "Describe this image", "assets/examples/test.png")
+
 if __name__ == "__main__":
     # test_openai_model()
     # test_google_model()