diff --git a/openllm-core/src/openllm_core/_configuration.py b/openllm-core/src/openllm_core/_configuration.py
index e250992f..bffa94e4 100644
--- a/openllm-core/src/openllm_core/_configuration.py
+++ b/openllm-core/src/openllm_core/_configuration.py
@@ -473,6 +473,8 @@ class LLMConfig(pydantic.BaseModel, abc.ABC):
   @overload
   def __getitem__(self, item: t.Literal['logits_processors']) -> t.Optional[t.List[LogitsProcessor]]: ...
   @overload
+  def __getitem__(self, item: t.Literal['seed']) -> t.Optional[int]: ...
+  @overload
   def __getitem__(self, item: t.Literal['max_new_tokens']) -> int: ...
   @overload
   def __getitem__(self, item: t.Literal['start_name']) -> str: ...
diff --git a/openllm-python/pyproject.toml b/openllm-python/pyproject.toml
index 95761dd0..0f18db20 100644
--- a/openllm-python/pyproject.toml
+++ b/openllm-python/pyproject.toml
@@ -124,7 +124,7 @@ playground = ["jupyter", "notebook", "ipython", "jupytext", "nbformat"]
 qwen = ["cpm-kernels", "tiktoken"]
 stablelm = ["cpm-kernels", "tiktoken"]
 starcoder = ["bitsandbytes"]
-vllm = ["vllm==0.4.0"]
+vllm = ["vllm==0.4.2"]
 yi = ["bitsandbytes"]
 
 [tool.hatch.version]
diff --git a/openllm-python/src/_openllm_tiny/_entrypoint.py b/openllm-python/src/_openllm_tiny/_entrypoint.py
index bbedc1f2..006edd67 100644
--- a/openllm-python/src/_openllm_tiny/_entrypoint.py
+++ b/openllm-python/src/_openllm_tiny/_entrypoint.py
@@ -289,7 +289,7 @@ def construct_python_options(llm_config, llm_fs):
 
   # TODO: Add this line back once 0.5 is out, for now depends on OPENLLM_DEV_BUILD
   # packages = ['scipy', 'bentoml[tracing]>=1.2.8', 'openllm[vllm]>0.4', 'vllm>=0.3']
-  packages = ['scipy', 'bentoml[tracing]>=1.2.8', 'vllm>=0.3', 'flash-attn']
+  packages = ['scipy', 'bentoml[tracing]>=1.2.8', 'vllm==0.4.2']
   if llm_config['requirements'] is not None:
     packages.extend(llm_config['requirements'])
   built_wheels = [build_editable(llm_fs.getsyspath('/'), p) for p in ('openllm_core', 'openllm_client', 'openllm')]
diff --git a/tools/dependencies.py b/tools/dependencies.py
index f80aec3b..ad47faaa 100755
--- a/tools/dependencies.py
+++ b/tools/dependencies.py
@@ -154,7 +154,7 @@ PLAYGROUND_DEPS = ['jupyter', 'notebook', 'ipython', 'jupytext', 'nbformat']
 GGML_DEPS = ['ctransformers']
 AWQ_DEPS = ['autoawq']
 GPTQ_DEPS = ['auto-gptq[triton]>=0.4.2']
-VLLM_DEPS = ['vllm==0.4.0']
+VLLM_DEPS = ['vllm==0.4.2']
 
 _base_requirements: dict[str, t.Any] = {
   inflection.dasherize(name): config_cls()['requirements']
@@ -290,7 +290,7 @@ def main(args) -> int:
     Dependencies(name='openllm-client', lower_constraint=release_version),
     Dependencies(name='openllm-core', lower_constraint=release_version),
     Dependencies(name='safetensors'),
-    Dependencies(name='vllm', lower_constraint='0.4.0'),
+    Dependencies(name='vllm', lower_constraint='0.4.2'),
     Dependencies(name='optimum', lower_constraint='1.12.0'),
     Dependencies(name='accelerate'),
     Dependencies(name='ghapi'),