fix: handle missing prompt_tokens_details in OpenAI-compatible APIs

Add checks for the existence of prompt_tokens_details in response objects
before attempting to access it. This fixes an issue when using third-party
OpenAI-compatible APIs that don't include cache-related information.

Fixes #82.

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <[email protected]>

Author: rishsriv

defog/llm/utils.py

@@ -753,13 +753,15 @@ async def _process_openai_response(
     if tools and len(tools) > 0:
         consecutive_exceptions = 0
         while True:
-            total_input_tokens += (
-                response.usage.prompt_tokens
-                - response.usage.prompt_tokens_details.cached_tokens
-            )
-            total_cached_input_tokens += (
-                response.usage.prompt_tokens_details.cached_tokens
-            )
+            # Check if prompt_tokens_details exists in the response
+            if hasattr(response.usage, 'prompt_tokens_details') and response.usage.prompt_tokens_details is not None:
+                cached_tokens = response.usage.prompt_tokens_details.cached_tokens
+                total_input_tokens += response.usage.prompt_tokens - cached_tokens
+                total_cached_input_tokens += cached_tokens
+            else:
+                # If prompt_tokens_details doesn't exist, assume all tokens are uncached
+                total_input_tokens += response.usage.prompt_tokens
+            
             total_output_tokens += response.usage.completion_tokens
             message = response.choices[0].message
             if message.tool_calls:
@@ -872,10 +874,15 @@ async def _process_openai_response(
             content = response.choices[0].message.content
 
     usage = response.usage
-    total_cached_input_tokens += usage.prompt_tokens_details.cached_tokens
-    total_input_tokens += (
-        usage.prompt_tokens - usage.prompt_tokens_details.cached_tokens
-    )
+    # Check if prompt_tokens_details exists in the response
+    if hasattr(usage, 'prompt_tokens_details') and usage.prompt_tokens_details is not None:
+        cached_tokens = usage.prompt_tokens_details.cached_tokens
+        total_cached_input_tokens += cached_tokens
+        total_input_tokens += usage.prompt_tokens - cached_tokens
+    else:
+        # If prompt_tokens_details doesn't exist, assume all tokens are uncached
+        total_input_tokens += usage.prompt_tokens
+    
     total_output_tokens += usage.completion_tokens
     return (
         content,