diff --git a/litellm/litellm_core_utils/exception_mapping_utils.py b/litellm/litellm_core_utils/exception_mapping_utils.py index 85c232ad3f4..f1195802fc5 100644 --- a/litellm/litellm_core_utils/exception_mapping_utils.py +++ b/litellm/litellm_core_utils/exception_mapping_utils.py @@ -107,6 +107,7 @@ class ExceptionCheckers: "exceeds the available context size", # llama.cpp/Lemonade "exceeds the maximum number of tokens allowed", # Gemini '"code": "context_length_exceeded"', + "prompt is too long", ] for substring in known_exception_substrings: if substring in _error_str_lowercase: diff --git a/tests/unit/litellm_core_utils/test_exception_mapping_utils.py b/tests/unit/litellm_core_utils/test_exception_mapping_utils.py index 2b1e4a5eed3..1b00a32d31d 100644 --- a/tests/unit/litellm_core_utils/test_exception_mapping_utils.py +++ b/tests/unit/litellm_core_utils/test_exception_mapping_utils.py @@ -603,6 +603,26 @@ def test_anthropic_context_window_error_mapping(error_message): assert excinfo.value.llm_provider == "anthropic" +@pytest.mark.parametrize("upstream_status_code", [400, 500]) +def test_vertex_ai_anthropic_prompt_too_long_maps_to_context_window_exceeded(upstream_status_code: int) -> None: + original_exception: Final = VertexAIError( + status_code=upstream_status_code, + message='{"type":"error","error":{"type":"invalid_request_error",' + '"message":"prompt is too long: 210123 tokens > 200000 maximum"}}', + ) + + with pytest.raises(litellm.ContextWindowExceededError) as excinfo: + exception_type( + model="claude-opus-5-5", + original_exception=original_exception, + custom_llm_provider="vertex_ai", + ) + + assert excinfo.value.status_code == 400 + assert excinfo.value.llm_provider == "vertex_ai" + assert excinfo.value.model == "claude-opus-5-5" + + # Test cases for Vertex AI RateLimitError mapping # As per https://github.com/BerriAI/litellm/issues/16189 vertex_rate_limit_test_cases = [