diff --git a/vllm/entrypoints/llm.py b/vllm/entrypoints/llm.py
index 892e5035ab..349091f4b7 100644
--- a/vllm/entrypoints/llm.py
+++ b/vllm/entrypoints/llm.py
@@ -898,6 +898,12 @@ class LLM(BeamSearchOfflineMixin, PoolingOfflineMixin, OfflineInferenceMixin):
     def finish_weight_update(self) -> None:
         """Finish the current weight update."""
         self.llm_engine.collective_rpc("finish_weight_update")
+        # Invalidate cached state computed with the old weights so it isn't
+        # reused for subsequent requests:
+        # - prefix cache: KV blocks computed with the old weights
+        # - encoder cache: multimodal embeddings keyed only by mm_hash
+        self.llm_engine.reset_prefix_cache()
+        self.llm_engine.reset_encoder_cache()
 
     def __repr__(self) -> str:
         """Return a transformers-style hierarchical view of the model."""
diff --git a/vllm/v1/engine/async_llm.py b/vllm/v1/engine/async_llm.py
index 419e15163a..26b3f53d2c 100644
--- a/vllm/v1/engine/async_llm.py
+++ b/vllm/v1/engine/async_llm.py
@@ -1109,3 +1109,9 @@ class AsyncLLM(EngineClient):
     async def finish_weight_update(self) -> None:
         """Finish the current weight update."""
         await self.collective_rpc("finish_weight_update")
+        # Invalidate cached state computed with the old weights so it isn't
+        # reused for subsequent requests:
+        # - prefix cache: KV blocks computed with the old weights
+        # - encoder cache: multimodal embeddings keyed only by mm_hash
+        await self.reset_prefix_cache()
+        await self.reset_encoder_cache()