fix(vllm-cpp): pin MLX 0.32.0, the version the engine builds against (#12734)

* chore(vllm-cpp): bump MLX pin to 0.32.4

The macOS runner's pip index no longer offers mlx==0.29.4 (it lists only
0.32.4), so the darwin vllm-cpp build fails at the MLX venv step. Move the
pin to 0.32.4 and update the README default to match.

Not verified on macOS: the Makefile was only dry-run on Linux.

Assisted-by: Claude Code:claude-sonnet-5-5
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

* fix(vllm-cpp): pin MLX 0.32.0, the version the engine builds against

The macOS build fails against mlx 0.32.4:

  metal_mlx_provider.mm:147:5: error: no matching member function
  for call to 'set_data'

The vllm.cpp Metal provider calls the five-argument
array::set_data(Buffer, size, strides, flags, Deleter). The 0.32.4
headers only declare set_data(allocator::Data) and the Data overload
with size, strides, flags and offset; 0.32.0 still has the Buffer one.

The engine's own CI installs 'mlx==0.32.0' (ci.yml line 1798,
release.yml line 360), so pin that and update the README default.

Not verified on macOS: only the Makefile dry-run and the wheel
headers were checked on Linux.

Assisted-by: Claude Code:claude-sonnet-5-5
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

* fix(vllm-cpp): install the MLX 0.32.0 wheel by URL and hash

The macOS runner's pip index offers only mlx 0.32.4, so
'pip install mlx==0.32.0' fails with "Could not find a version that
satisfies the requirement". 0.32.4 cannot be used: its headers lack the
five-argument array::set_data the engine calls.

Install the 0.32.0 wheel directly by URL with a sha256 check. The wheel
is mlx-metal (py3-none-macosx_14_0_arm64) from PyPI, sha256
5b64b20ac24b0c401f489de01e8209edc4d372125201f19314e6f39e385322aa. It
holds mlx/include, mlx/lib/libmlx.dylib and mlx/lib/mlx.metallib, which
is all the build and package.sh use, and it is not tied to a CPython
version. The URL and hash are Makefile variables; the stamp stays keyed
on MLX_VERSION.

Not verified on macOS: only the Makefile dry-run, the wheel contents and
its sha256 were checked on Linux.

Assisted-by: Claude Code:claude-sonnet-5-5
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

---------

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
Co-authored-by: Ettore Di Giacinto <mudler@localai.io>

backend/go/vllm-cpp/Makefile

@@ -30,7 +30,15 @@ VLLM_CPP_VERSION?=bf7b654ae2d0cf49028291192a78bd92f2843106
 # decode makes ~112 calls per TOKEN. Ungated it does both; gated it does only the
 # good half. So if this pin is ever moved BACKWARDS, this default must go with it.
 VLLM_CPP_MLX?=on
-MLX_VERSION?=0.29.4
+MLX_VERSION?=0.32.0
+# The macOS runners' pip index offers only the newest MLX (0.32.4), which lacks
+# the array::set_data overload the engine calls, so the pinned wheel is fetched
+# by URL and verified by sha256. This is the mlx-metal distribution: it holds
+# mlx/include, mlx/lib/libmlx.dylib and mlx/lib/mlx.metallib, the only parts the
+# build uses, and unlike the `mlx` wheel it is not tied to one CPython version.
+# Bumping MLX_VERSION means bumping both values (PyPI JSON: /pypi/mlx-metal/<v>/json).
+MLX_WHEEL_URL?=https://files.pythonhosted.org/packages/d6/ef/d74ae99cfe9ddb59fd08abd14f47754c3d199291a93756903fa595b31b8a/mlx_metal-0.32.0-py3-none-macosx_14_0_arm64.whl
+MLX_WHEEL_SHA256?=5b64b20ac24b0c401f489de01e8209edc4d372125201f19314e6f39e385322aa
 MLX_VENV?=$(abspath ./mlx-venv)
 # Resolved lazily (recursive `=`, not `:=`): the glob only matches once the venv
 # target has run, and the interpreter version in the path varies per runner.
@@ -153,14 +161,14 @@ $(MLX_STAMP):
 	@if [ ! -x "$(MLX_VENV)/bin/pip" ]; then \
 		python3 -m venv "$(MLX_VENV)" || { echo "vllm-cpp: python3 with venv is required to build the MLX provider; pass VLLM_CPP_MLX=off to build Metal without it" >&2; exit 1; }; \
 	fi
-	"$(MLX_VENV)"/bin/pip install --quiet --disable-pip-version-check "mlx==$(MLX_VERSION)"
+	"$(MLX_VENV)"/bin/pip install --quiet --disable-pip-version-check "mlx-metal @ $(MLX_WHEEL_URL)#sha256=$(MLX_WHEEL_SHA256)"
 	@# Resolved in the SHELL, not by $(MLX_ROOT): make expands a whole recipe
 	@# before running its first line, so the glob would still be unmatched here.
 	@# Every later use (the cmake args, package.sh) expands after this target has
 	@# completed, where $(MLX_ROOT) does resolve.
 	@root=$$(echo "$(MLX_VENV)"/lib/python*/site-packages/mlx); \
 	test -f "$$root/lib/libmlx.dylib" -a -f "$$root/include/mlx/array.h" || \
-		{ echo "vllm-cpp: mlx==$(MLX_VERSION) did not provide lib/libmlx.dylib + include/mlx/array.h under $$root" >&2; exit 1; }
+		{ echo "vllm-cpp: mlx-metal $(MLX_VERSION) did not provide lib/libmlx.dylib + include/mlx/array.h under $$root" >&2; exit 1; }
 	touch $@
 else
 MLX_STAMP=

backend/go/vllm-cpp/README.md

@@ -181,9 +181,12 @@ Build knobs:
 
 - `VLLM_CPP_MLX=off` builds Metal without the provider: ~124 MB smaller, and
   96.4% of MLX-LM instead of 99.1%.
-- `MLX_VERSION` pins the wheel (default `0.29.4`). MLX is consumed as the
-  prebuilt pip wheel because building it from source needs `xcrun metal`, i.e. a
-  full Xcode the macOS runners do not have.
+- `MLX_VERSION` names the wheel (default `0.32.0`). MLX is consumed as the
+  prebuilt `mlx-metal` wheel because building it from source needs `xcrun metal`,
+  i.e. a full Xcode the macOS runners do not have. The default wheel is installed
+  by URL and sha256 (`MLX_WHEEL_URL`, `MLX_WHEEL_SHA256`), because the runners'
+  pip index offers only 0.32.4, whose headers lack the `array::set_data` overload
+  the engine calls. Change all three together when you bump MLX.
 
 Packaging vendors `libmlx.dylib`, `mlx.metallib` and MLX's MIT license into
 `package/lib/`, and rewrites `libvllm.dylib`'s rpath to `@loader_path/lib`