From 51f0acda7960871f9fdc81d79481b18bee957ea8 Mon Sep 17 00:00:00 2001
From: Cyrus Leung <tlleungac@connect.ust.hk>
Date: Wed, 18 Mar 2026 03:44:52 +0800
Subject: [PATCH] [Model] Remove unused `handle_oov_mm_token` (#37321)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
---
 .../models/multimodal/processing/test_qwen2_5_omni_embed.py | 6 ++++--
 vllm/model_executor/models/fireredasr2.py                   | 1 -
 vllm/model_executor/models/kimi_audio.py                    | 1 -
 3 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/tests/models/multimodal/processing/test_qwen2_5_omni_embed.py b/tests/models/multimodal/processing/test_qwen2_5_omni_embed.py
index 5001b98b6d..4eb4d03bfe 100644
--- a/tests/models/multimodal/processing/test_qwen2_5_omni_embed.py
+++ b/tests/models/multimodal/processing/test_qwen2_5_omni_embed.py
@@ -185,14 +185,16 @@ def make_mock_model(hidden: int = 8):
 
     # super().embed_input_ids → use SupportsMultiModal.embed_input_ids
     def fake_super_embed(
-        ids, mm_embs=None, *, is_multimodal=None, handle_oov_mm_token=False
+        ids,
+        mm_embs=None,
+        *,
+        is_multimodal=None,
     ):
         return SupportsMultiModal.embed_input_ids(
             model,
             ids,
             mm_embs,
             is_multimodal=is_multimodal,
-            handle_oov_mm_token=handle_oov_mm_token,
         )
 
     # Bind embed_input_ids as the real method
diff --git a/vllm/model_executor/models/fireredasr2.py b/vllm/model_executor/models/fireredasr2.py
index 5d6c684546..0aae13997c 100644
--- a/vllm/model_executor/models/fireredasr2.py
+++ b/vllm/model_executor/models/fireredasr2.py
@@ -793,7 +793,6 @@ class FireRedASR2ForConditionalGeneration(
         multimodal_embeddings: MultiModalEmbeddings | None = None,
         *,
         is_multimodal: torch.Tensor | None = None,
-        handle_oov_mm_token: bool = False,
     ) -> torch.Tensor:
         inputs_embeds = self.model.decoder.embed_input_ids(input_ids)
 
diff --git a/vllm/model_executor/models/kimi_audio.py b/vllm/model_executor/models/kimi_audio.py
index 36d22d867f..651144683f 100644
--- a/vllm/model_executor/models/kimi_audio.py
+++ b/vllm/model_executor/models/kimi_audio.py
@@ -514,7 +514,6 @@ class KimiAudioForConditionalGeneration(
         multimodal_embeddings: tuple[torch.Tensor, ...] | None = None,
         *,
         is_multimodal: torch.Tensor | None = None,
-        handle_oov_mm_token: bool = False,
     ) -> torch.Tensor:
         """Embed input IDs and fuse with audio embeddings.