Merge pull request #165 from ridgesai/hobbleabbas-patch-7

Increase load balancing for glm4.5fp8 to targon

Author: hobbleabbas

proxy/providers/inference_manager.py

@@ -38,7 +38,7 @@ def _find_provider(self, model: str) -> InferenceProvider:
         if (
             self.targon.supports_model(model)
             and self.targon.is_available()
-            and random.random() < 0.25
+            and random.random() < 0.5
         ):
             return self.targon
 
@@ -153,4 +153,4 @@ async def inference(
                 return {"error": f"{primary_provider.name} error: {response_text} | {fallback_provider.name} fallback also failed: {fallback_response}"}
 
         # No fallback available or model doesn't support fallback
-        return {"error": f"Error in inference request: {response_text}"} 
+        return {"error": f"Error in inference request: {response_text}"}