diff --git a/litellm/llms/oci/chat/cohere.py b/litellm/llms/oci/chat/cohere.py
index 7805f17c74c..7e380cf0e7f 100644
--- a/litellm/llms/oci/chat/cohere.py
+++ b/litellm/llms/oci/chat/cohere.py
@@ -10,6 +10,7 @@ import datetime
 import json
 from typing import Any, Dict, List, Optional
 
+import httpx
 from pydantic import ValidationError
 
 from litellm.llms.oci.chat.generic import _synthesize_oci_tool_call_id
@@ -202,9 +203,16 @@ def handle_cohere_response(
     json_response: dict,
     model: str,
     model_response: ModelResponse,
+    raw_response: httpx.Response,
 ) -> ModelResponse:
     """Parse a non-streaming Cohere OCI response into a LiteLLM ModelResponse."""
-    cohere_response = CohereChatResult(**json_response)
+    try:
+        cohere_response = CohereChatResult(**json_response)
+    except (TypeError, ValidationError) as e:
+        raise OCIError(
+            message=f"Response cannot be casted to CohereChatResult: {str(e)}",
+            status_code=raw_response.status_code,
+        )
 
     model_response.model = model
     model_response.created = int(datetime.datetime.now().timestamp())
diff --git a/litellm/llms/oci/chat/transformation.py b/litellm/llms/oci/chat/transformation.py
index a06c0ebdd3f..2b928220fb6 100644
--- a/litellm/llms/oci/chat/transformation.py
+++ b/litellm/llms/oci/chat/transformation.py
@@ -543,7 +543,7 @@ class OCIChatConfig(BaseConfig):
         vendor = get_vendor_from_model(model)
         if vendor == OCIVendors.COHERE:
             model_response = handle_cohere_response(
-                response_json, model, model_response
+                response_json, model, model_response, raw_response
             )
         else:
             model_response = handle_generic_response(
diff --git a/tests/test_litellm/llms/oci/test_oci_coverage_boost.py b/tests/test_litellm/llms/oci/test_oci_coverage_boost.py
index be3de91ba3d..c8934e26456 100644
--- a/tests/test_litellm/llms/oci/test_oci_coverage_boost.py
+++ b/tests/test_litellm/llms/oci/test_oci_coverage_boost.py
@@ -532,10 +532,13 @@ _COHERE_RESPONSE_JSON = {
 }
 
 
+_COHERE_RAW_RESPONSE = httpx.Response(200, request=httpx.Request("POST", "https://oci"))
+
+
 def test_handle_cohere_response_complete():
     model_response = ModelResponse()
     result = handle_cohere_response(
-        _COHERE_RESPONSE_JSON, _COHERE_MODEL, model_response
+        _COHERE_RESPONSE_JSON, _COHERE_MODEL, model_response, _COHERE_RAW_RESPONSE
     )
     assert result.choices[0].finish_reason == "stop"
     assert result.choices[0].message["content"] == "Hello from Cohere!"
@@ -551,7 +554,9 @@ def test_handle_cohere_response_max_tokens():
         },
     }
     model_response = ModelResponse()
-    result = handle_cohere_response(resp, _COHERE_MODEL, model_response)
+    result = handle_cohere_response(
+        resp, _COHERE_MODEL, model_response, _COHERE_RAW_RESPONSE
+    )
     assert result.choices[0].finish_reason == "length"
 
 
@@ -565,7 +570,9 @@ def test_handle_cohere_response_tool_call():
         },
     }
     model_response = ModelResponse()
-    result = handle_cohere_response(resp, _COHERE_MODEL, model_response)
+    result = handle_cohere_response(
+        resp, _COHERE_MODEL, model_response, _COHERE_RAW_RESPONSE
+    )
     assert result.choices[0].finish_reason == "tool_calls"
     tool_calls = result.choices[0].message["tool_calls"]
     assert tool_calls is not None
@@ -582,12 +589,23 @@ def test_handle_cohere_response_missing_usage():
         },
     }
     model_response = ModelResponse()
-    result = handle_cohere_response(resp, _COHERE_MODEL, model_response)
+    result = handle_cohere_response(
+        resp, _COHERE_MODEL, model_response, _COHERE_RAW_RESPONSE
+    )
     assert result.usage.prompt_tokens == 0
     assert result.usage.completion_tokens == 0
     assert result.usage.total_tokens == 0
 
 
+def test_handle_cohere_response_malformed_raises_oci_error():
+    bad_json = {"chatResponse": {"apiFormat": "COHERE"}}
+    raw = httpx.Response(502, request=httpx.Request("POST", "https://oci"))
+    model_response = ModelResponse()
+    with pytest.raises(OCIError) as exc_info:
+        handle_cohere_response(bad_json, _COHERE_MODEL, model_response, raw)
+    assert exc_info.value.status_code == 502
+
+
 # ===========================================================================
 # cohere.py — handle_cohere_stream_chunk
 # ===========================================================================