fix(rerank): stamp a fresh response id when Voyage, watsonx, or Fireworks omit one

litellm/llms/fireworks_ai/rerank/transformation.py

@@ -250,8 +250,7 @@ class FireworksAIRerankConfig(FireworksAIMixin, BaseRerankConfig):
 
             rerank_results.append(rerank_result)
 
-        # Use model name as id if no id is provided
-        response_id: Final = raw_response_json.get("id") or raw_response_json.get("model") or str(uuid.uuid4())
+        response_id: Final = raw_response_json.get("id") or str(uuid.uuid4())
 
         return RerankResponse(
             id=response_id,

litellm/llms/voyage/rerank/transformation.py

@@ -9,6 +9,7 @@ from typing import Any, Final
 
 import httpx
 
+from litellm._uuid import uuid
 from litellm.llms.base_llm.chat.transformation import LiteLLMLoggingObj
 from litellm.llms.base_llm.rerank.transformation import BaseRerankConfig
 from litellm.secret_managers.main import get_secret_str
@@ -127,7 +128,7 @@ class VoyageRerankConfig(BaseRerankConfig):
         rerank_meta: Final = RerankResponseMeta(billed_units=_billed_units, tokens=_tokens)
 
         return RerankResponse(
-            id=_json_response.get("id", f"voyage-rerank-{model}"),
+            id=_json_response.get("id") or str(uuid.uuid4()),
             results=transformed_results,
             meta=rerank_meta,
         )

litellm/llms/watsonx/rerank/transformation.py

@@ -191,7 +191,7 @@ class IBMWatsonXRerankConfig(IBMWatsonXMixin, BaseRerankConfig):
 
             transformed_results.append(transformed_result)
 
-        response_id: Final = raw_response_json.get("id") or raw_response_json.get("model_id") or str(uuid.uuid4())
+        response_id: Final = raw_response_json.get("id") or str(uuid.uuid4())
 
         # Extract usage information
         _tokens: Final = RerankTokens(

tests/test_litellm/llms/fireworks_ai/rerank/test_fireworks_ai_rerank_transformation.py

@@ -3,6 +3,7 @@ Tests for Fireworks AI rerank transformation functionality.
 """
 
 import json
+import uuid
 from unittest.mock import MagicMock
 
 import httpx
@@ -181,8 +182,7 @@ class TestFireworksAIRerankTransform:
         )
 
         # Verify response structure
-        # Fireworks AI doesn't return "id", so it uses "model" as the id
-        assert result.id == "accounts/fireworks/models/qwen3-reranker-8b"
+        assert uuid.UUID(result.id).version == 4
         assert len(result.results) == 2
         assert result.results[0]["index"] == 0
         assert result.results[0]["relevance_score"] == 0.95
@@ -229,16 +229,14 @@ class TestFireworksAIRerankTransform:
             logging_obj=mock_logging,
         )
 
-        # Fireworks AI doesn't return "id", so it uses "model" as the id
-        assert result.id == "accounts/fireworks/models/qwen3-reranker-8b"
+        assert uuid.UUID(result.id).version == 4
         assert len(result.results) == 2
         assert result.results[0]["index"] == 0
         assert result.results[0]["relevance_score"] == 0.95
         # Document should not be present
         assert "document" not in result.results[0]
 
-    def test_transform_rerank_response_missing_id(self):
-        """Test response transformation when id is missing (should use model name or generate UUID)."""
+    def test_transform_rerank_response_missing_id_stamps_a_fresh_id_per_call(self):
         response_data = {
             "object": "list",
             "model": "accounts/fireworks/models/qwen3-reranker-8b",
@@ -248,23 +246,22 @@ class TestFireworksAIRerankTransform:
             "usage": {"total_tokens": 10},
         }
 
-        mock_response = MagicMock(spec=httpx.Response)
-        mock_response.json.return_value = response_data
-        mock_response.status_code = 200
-        mock_response.headers = {}
+        def transform() -> str:
+            mock_response = MagicMock(spec=httpx.Response)
+            mock_response.json.return_value = response_data
+            mock_response.status_code = 200
+            mock_response.headers = {}
+            return self.config.transform_rerank_response(
+                model=self.model,
+                raw_response=mock_response,
+                model_response=RerankResponse(),
+                logging_obj=MagicMock(),
+            ).id
 
-        mock_logging = MagicMock()
-        model_response = RerankResponse()
+        first, second = transform(), transform()
 
-        result = self.config.transform_rerank_response(
-            model=self.model,
-            raw_response=mock_response,
-            model_response=model_response,
-            logging_obj=mock_logging,
-        )
-
-        # Should use model name when id is missing
-        assert result.id == "accounts/fireworks/models/qwen3-reranker-8b"
+        assert first != second
+        assert "accounts/fireworks/models/qwen3-reranker-8b" not in (first, second)
 
     def test_transform_rerank_response_missing_results(self):
         """Test that missing results raises ValueError."""

tests/test_litellm/llms/voyage/rerank/test_voyage_rerank_transformation.py

@@ -3,6 +3,7 @@ Tests for Voyage AI rerank transformation functionality.
 """
 
 import json
+import uuid
 from unittest.mock import MagicMock, patch
 
 import httpx
@@ -258,6 +259,33 @@ class TestVoyageRerankTransform:
 
         assert "Failed to parse response" in str(exc_info.value)
 
+    def test_transform_rerank_response_without_id_stamps_a_fresh_id_per_call(self):
+        response_data = {
+            "object": "list",
+            "data": [{"relevance_score": 0.5, "index": 0}],
+            "model": "rerank-2.5",
+            "usage": {"total_tokens": 10},
+        }
+
+        def transform() -> str:
+            mock_response = MagicMock(spec=httpx.Response)
+            mock_response.json.return_value = response_data
+            mock_response.status_code = 200
+            mock_response.text = json.dumps(response_data)
+            mock_response.headers = {}
+            return self.config.transform_rerank_response(
+                model=self.model,
+                raw_response=mock_response,
+                model_response=RerankResponse(),
+                logging_obj=MagicMock(),
+            ).id
+
+        first, second = transform(), transform()
+
+        assert uuid.UUID(first).version == 4
+        assert first != second
+        assert f"voyage-rerank-{self.model}" not in (first, second)
+
     def test_get_supported_cohere_rerank_params(self):
         """Test getting supported parameters for Voyage AI rerank."""
         supported_params = self.config.get_supported_cohere_rerank_params(self.model)

tests/test_litellm/llms/watsonx/rerank/test_watsonx_rerank.py

@@ -120,9 +120,7 @@ class TestIBMWatsonXRerankTransform:
             logging_obj=mock_logging,
         )
 
-        # Verify response structure
-        # IBM watsonx.ai doesn't return "id", so it uses "model" as the id
-        assert result.id == "watsonx/cross-encoder/ms-marco-minilm-l-12-v2"
+        assert uuid.UUID(result.id).version == 4
         assert len(result.results) == 2
         assert result.results[0]["index"] == 0
         assert result.results[0]["relevance_score"] == 6.53515625
@@ -172,9 +170,7 @@ class TestIBMWatsonXRerankTransform:
             logging_obj=mock_logging,
         )
 
-        # Verify response structure
-        # IBM watsonx.ai doesn't return "id", so it uses "model" as the id
-        assert result.id == "watsonx/cross-encoder/ms-marco-minilm-l-12-v2"
+        assert uuid.UUID(result.id).version == 4
         assert len(result.results) == 2
 
         assert result.results[0]["index"] == 0
@@ -231,6 +227,30 @@ class TestIBMWatsonXRerankTransform:
                 logging_obj=mock_logging,
             )
 
+    def test_transform_rerank_response_without_id_stamps_a_fresh_id_per_call(self):
+        response_data = {
+            "model_id": self.model,
+            "results": [{"index": 0, "score": 1.5}],
+            "input_token_count": 12,
+        }
+
+        def transform() -> str:
+            mock_response = MagicMock(spec=httpx.Response)
+            mock_response.json.return_value = response_data
+            mock_response.status_code = 200
+            mock_response.headers = {}
+            return self.config.transform_rerank_response(
+                model=self.model,
+                raw_response=mock_response,
+                model_response=RerankResponse(),
+                logging_obj=MagicMock(),
+            ).id
+
+        first, second = transform(), transform()
+
+        assert first != second
+        assert self.model not in (first, second)
+
     def test_get_supported_cohere_rerank_params(self):
         """Test getting supported parameters for IBM watsonx.ai rerank."""
         supported_params = self.config.get_supported_cohere_rerank_params(self.model)