feat: Enable output schema with tools for LiteLlm models

LiteLlm provides built-in handling for tool and response format compatibility across different providers, allowing output schemas to be used reliably with tools for any LiteLlm instance

Close #3969

Co-authored-by: George Weale <gweale@google.com>
PiperOrigin-RevId: 877471153

src/google/adk/utils/output_schema_utils.py

@@ -30,6 +30,17 @@ from .variant_utils import GoogleLLMVariant
 
 def can_use_output_schema_with_tools(model: Union[str, BaseLlm]) -> bool:
   """Returns True if output schema with tools is supported."""
+  # LiteLLM handles tools + response_format compatibility per-provider:
+  # - Providers with native support (OpenAI, Azure): both passed directly
+  # - Providers without (Fireworks): auto-converted to json_tool_call +
+  #   tool_choice enforcement
+  # This is strictly more reliable than the SetModelResponseTool
+  # prompt-based workaround.
+  from ..models.lite_llm import LiteLlm
+
+  if isinstance(model, LiteLlm):
+    return True
+
   model_string = model if isinstance(model, str) else model.model
 
   return (

tests/unittests/utils/test_output_schema_utils.py

@@ -15,6 +15,7 @@
 
 from google.adk.models.anthropic_llm import Claude
 from google.adk.models.google_llm import Gemini
+from google.adk.models.lite_llm import LiteLlm
 from google.adk.utils.output_schema_utils import can_use_output_schema_with_tools
 import pytest
 
@@ -37,6 +38,11 @@ import pytest
         (Claude(model="claude-3.7-sonnet"), "1", False),
         (Claude(model="claude-3.7-sonnet"), "0", False),
         (Claude(model="claude-3.7-sonnet"), None, False),
+        (LiteLlm(model="openai/gpt-4o"), "1", True),
+        (LiteLlm(model="openai/gpt-4o"), "0", True),
+        (LiteLlm(model="openai/gpt-4o"), None, True),
+        (LiteLlm(model="anthropic/claude-3.7-sonnet"), None, True),
+        (LiteLlm(model="fireworks_ai/llama-v3p1-70b"), None, True),
     ],
 )
 def test_can_use_output_schema_with_tools(