feat: extended /v1/models endpoint, now it returns with fallbacks on demand

litellm/proxy/auth/model_checks.py

@@ -6,6 +6,7 @@
 from litellm._logging import verbose_proxy_logger
 from litellm.proxy._types import SpecialModelNames, UserAPIKeyAuth
 from litellm.router import Router
+from litellm.router_utils.fallback_event_handlers import get_fallback_model_group
 from litellm.types.router import LiteLLM_Params
 from litellm.utils import get_valid_models
 
@@ -52,7 +53,7 @@ def _get_models_from_access_groups(
         if model in model_access_groups:
             if (
                 not include_model_access_groups
-            ):  # remove access group, unless requested - e.g. when creating a key and trying to see list of models
+            ):  # remove access group, unless requested - e.g. when creating a key
                 idx_to_remove.append(idx)
             new_models.extend(model_access_groups[model])
 
@@ -104,7 +105,8 @@ def get_key_models(
     - List of model name strings
     - Empty list if no models set
     - If model_access_groups is provided, only return models that are in the access groups
-    - If include_model_access_groups is True, it includes the 'keys' of the model_access_groups in the response - {"beta-models": ["gpt-4", "claude-v1"]} -> returns 'beta-models'
+    - If include_model_access_groups is True, it includes the 'keys' of the model_access_groups
+      in the response - {"beta-models": ["gpt-4", "claude-v1"]} -> returns 'beta-models'
     """
     all_models: List[str] = []
     if len(user_api_key_dict.models) > 0:
@@ -287,3 +289,53 @@ def _get_wildcard_models(
         unique_models.remove(model)
 
     return all_wildcard_models
+
+
+def get_all_fallbacks(
+    model: str,
+    llm_router: Optional[Router] = None,
+    fallback_type: str = "general",
+) -> List[str]:
+    """
+    Get all fallbacks for a given model from the router's fallback configuration.
+
+    Args:
+        model: The model name to get fallbacks for
+        llm_router: The LiteLLM router instance
+        fallback_type: Type of fallback ("general", "context_window", "content_policy")
+
+    Returns:
+        List of fallback model names. Empty list if no fallbacks found.
+    """
+    if llm_router is None:
+        return []
+
+    # Get the appropriate fallback list based on type
+    fallbacks_config: list = []
+    if fallback_type == "general":
+        fallbacks_config = getattr(llm_router, "fallbacks", [])
+    elif fallback_type == "context_window":
+        fallbacks_config = getattr(llm_router, "context_window_fallbacks", [])
+    elif fallback_type == "content_policy":
+        fallbacks_config = getattr(llm_router, "content_policy_fallbacks", [])
+    else:
+        verbose_proxy_logger.warning(f"Unknown fallback_type: {fallback_type}")
+        return []
+
+    if not fallbacks_config:
+        return []
+
+    try:
+        # Use existing function to get fallback model group
+        fallback_model_group, _ = get_fallback_model_group(
+            fallbacks=fallbacks_config, 
+            model_group=model
+        )
+
+        if fallback_model_group is None:
+            return []
+
+        return fallback_model_group
+    except Exception as e:
+        verbose_proxy_logger.error(f"Error getting fallbacks for model {model}: {e}")
+        return []

litellm/proxy/proxy_server.py

@@ -167,6 +167,7 @@ def generate_feedback_box():
 from litellm.proxy.auth.handle_jwt import JWTHandler
 from litellm.proxy.auth.litellm_license import LicenseCheck
 from litellm.proxy.auth.model_checks import (
+    get_all_fallbacks,
     get_complete_model_list,
     get_key_models,
     get_mcp_server_ids,
@@ -3657,11 +3658,18 @@ async def model_list(
     team_id: Optional[str] = None,
     include_model_access_groups: Optional[bool] = False,
     only_model_access_groups: Optional[bool] = False,
+    include_metadata: Optional[bool] = False,
+    fallback_type: Optional[str] = None,
 ):
     """
     Use `/model/info` - to get detailed model information, example - pricing, mode, etc.
 
     This is just for compatibility with openai projects like aider.
+
+    Query Parameters:
+    - include_metadata: Include additional metadata in the response with fallback information
+    - fallback_type: Type of fallbacks to include ("general", "context_window", "content_policy")
+                    Defaults to "general" when include_metadata=true
     """
     global llm_model_list, general_settings, llm_router, prisma_client, user_api_key_cache, proxy_logging_obj
     all_models = []
@@ -3722,16 +3730,44 @@ async def model_list(
         only_model_access_groups=only_model_access_groups,
     )
 
+    # Build response data
+    model_data = []
+    for model in all_models:
+        model_info = {
+            "id": model,
+            "object": "model",
+            "created": DEFAULT_MODEL_CREATED_AT_TIME,
+            "owned_by": "openai",
+        }
+
+        # Add metadata if requested
+        if include_metadata:
+            metadata = {}
+
+            # Default fallback_type to "general" if include_metadata is true
+            effective_fallback_type = fallback_type if fallback_type is not None else "general"
+
+            # Validate fallback_type
+            valid_fallback_types = ["general", "context_window", "content_policy"]
+            if effective_fallback_type not in valid_fallback_types:
+                raise HTTPException(
+                    status_code=400,
+                    detail=f"Invalid fallback_type. Must be one of: {valid_fallback_types}"
+                )
+
+            fallbacks = get_all_fallbacks(
+                model=model,
+                llm_router=llm_router,
+                fallback_type=effective_fallback_type
+            )
+            metadata["fallbacks"] = fallbacks
+
+            model_info["metadata"] = metadata
+
+        model_data.append(model_info)
+
     return dict(
-        data=[
-            {
-                "id": model,
-                "object": "model",
-                "created": DEFAULT_MODEL_CREATED_AT_TIME,
-                "owned_by": "openai",
-            }
-            for model in all_models
-        ],
+        data=model_data,
         object="list",
     )