diff --git a/server/internal/proxy/convert/stream.go b/server/internal/proxy/convert/stream.go
index 77a779e..7630af1 100644
--- a/server/internal/proxy/convert/stream.go
+++ b/server/internal/proxy/convert/stream.go
@@ -412,9 +412,10 @@ func (t *responsesToChat) line(line []byte) []byte {
 
 type chatToResponses struct {
 	sseState
-	model string
-	usage any
-	done  bool
+	model      string
+	usage      any
+	finishSeen bool
+	done       bool
 }
 
 func newChatToResponses() *chatToResponses { return &chatToResponses{} }
@@ -425,6 +426,7 @@ func (t *chatToResponses) line(line []byte) []byte {
 		return nil
 	}
 	if done {
+		// 流结束兜底：finish 后 usage 未随块到达时在此补发 completed
 		if !t.done {
 			t.done = true
 			return eventLine("response.completed", map[string]any{
@@ -453,6 +455,9 @@ func (t *chatToResponses) line(line []byte) []byte {
 			finish, _ = c0["finish_reason"].(string)
 		}
 	}
+	if finish != "" {
+		t.finishSeen = true
+	}
 	var out [][]byte
 	if role, _ := delta["role"].(string); role == "assistant" {
 		out = append(out, eventLine("response.created", map[string]any{
@@ -465,7 +470,8 @@ func (t *chatToResponses) line(line []byte) []byte {
 			"type": "response.output_text.delta", "delta": content, "item_id": "msg_1", "output_index": 0, "content_index": 0,
 		}))
 	}
-	if finish != "" && !t.done {
+	// 上游 usage 块(choices 为空)通常晚于 finish_reason：此时再发 completed，携带 usage
+	if _, hasUsage := m["usage"]; hasUsage && t.finishSeen && !t.done {
 		t.done = true
 		out = append(out, eventLine("response.completed", map[string]any{
 			"type": "response.completed",