diff --git a/pkgs/development/python-modules/vllm/0005-drop-intel-reqs.patch b/pkgs/development/python-modules/vllm/0005-drop-intel-reqs.patch
index de5c9e8f0660..d6cee1f184b9 100644
--- a/pkgs/development/python-modules/vllm/0005-drop-intel-reqs.patch
+++ b/pkgs/development/python-modules/vllm/0005-drop-intel-reqs.patch
@@ -1,14 +1,13 @@
 diff --git a/requirements/cpu.txt b/requirements/cpu.txt
-index d11787df4..71575d707 100644
+index 21571be47..e91b378e2 100644
 --- a/requirements/cpu.txt
 +++ b/requirements/cpu.txt
-@@ -20,9 +20,6 @@ torchvision; platform_machine != "ppc64le" and platform_machine != "s390x"
- torchvision==0.23.0; platform_machine == "ppc64le"
- datasets # for benchmark scripts
+@@ -14,8 +14,5 @@ torchaudio; platform_machine != "s390x"
+ # required for the image processor of phi3v, this must be updated alongside torch
+ torchvision; platform_machine != "s390x"
  
 -# Intel Extension for PyTorch, only for x86_64 CPUs
 -intel-openmp==2024.2.1; platform_machine == "x86_64"
--intel_extension_for_pytorch==2.8.0; platform_machine == "x86_64"
- triton==3.2.0; platform_machine == "x86_64" # Triton is required for torch 2.6+cpu, as it is imported in torch.compile.
- 
+-
  # Use this to gather CPU info and optimize based on ARM Neoverse cores
+ py-cpuinfo; platform_machine == "aarch64"
diff --git a/pkgs/development/python-modules/vllm/default.nix b/pkgs/development/python-modules/vllm/default.nix
index 5097a3202cbf..3b569c9c4ef2 100644
--- a/pkgs/development/python-modules/vllm/default.nix
+++ b/pkgs/development/python-modules/vllm/default.nix
@@ -175,8 +175,8 @@ let
       name = "flash-attention-source";
       owner = "vllm-project";
       repo = "flash-attention";
-      rev = "58e0626a692f09241182582659e3bf8f16472659";
-      hash = "sha256-ewdZd7LuBKBV0y3AaGRWISJzjg6cu59D2OtgqoDjrbM=";
+      rev = "86f8f157cf82aa2342743752b97788922dd7de43";
+      hash = "sha256-+h43jMte/29kraNtPiloSQFfCay4W3NNIlzvs47ygyM=";
     };
 
     patches = [
@@ -304,7 +304,7 @@ in
 
 buildPythonPackage rec {
   pname = "vllm";
-  version = "0.11.2";
+  version = "0.12.0";
   pyproject = true;
 
   stdenv = torch.stdenv;
@@ -313,7 +313,7 @@ buildPythonPackage rec {
     owner = "vllm-project";
     repo = "vllm";
     tag = "v${version}";
-    hash = "sha256-DoSlkFmR3KKEtfSfdRB++0CZeeXgxmM3zZjONlxbe8U=";
+    hash = "sha256-ioAgZZbMv99UudaHtb3KQFAdjJv9GqeNDXDAqQOIMN8=";
   };
 
   patches = [