diff --git a/chat.go b/chat.go
index c8a3e81b..2027e9e4 100644
--- a/chat.go
+++ b/chat.go
@@ -389,7 +389,7 @@ type ChatCompletionChoice struct {
 type ChatCompletionResponse struct {
 	ID                  string                 `json:"id"`
 	Object              string                 `json:"object"`
-	Created             int64                  `json:"created"`
+	Created             interface{}            `json:"created,omitempty"`
 	Model               string                 `json:"model"`
 	Choices             []ChatCompletionChoice `json:"choices"`
 	Usage               Usage                  `json:"usage"`
diff --git a/chat_test.go b/chat_test.go
index 514706c9..6e8d8dc4 100644
--- a/chat_test.go
+++ b/chat_test.go
@@ -773,7 +773,7 @@ func handleChatCompletionEndpoint(w http.ResponseWriter, r *http.Request) {
 	res := openai.ChatCompletionResponse{
 		ID:      strconv.Itoa(int(time.Now().Unix())),
 		Object:  "test-object",
-		Created: time.Now().Unix(),
+		Created: time.Now().Format(time.RFC3339),
 		// would be nice to validate Model during testing, but
 		// this may not be possible with how much upkeep
 		// would be required / wouldn't make much sense