# `vllm_mlx.vllm_platform`
MLX Platform implementation for vLLM.
[View the complete module source at #L1-L351](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L1-L351).
## API details
Each callable below includes its exact signature, type annotations, inputs, defaults, return contract, documented exceptions, implementation source, and parsed docstring sections when the source provides them.
::: vllm_mlx.vllm_platform
options:
members:
- logger
- _get_apple_chip_name
- _get_unified_memory_size
- _is_mlx_available
- _is_apple_silicon
- MLXPlatform
filters: []
show_if_no_docstring: true
## Complete contract reference
Expand any definition for its exact inputs, annotations, defaults, return contract, directly raised exceptions, source-grounded behavior, and immutable line link. This section includes private and nested definitions that ordinary API generators omit.
vllm_mlx.vllm_platform._get_apple_chip_name · function
```python
vllm_mlx.vllm_platform._get_apple_chip_name() -> str
```
Get the name of the Apple Silicon chip.
**Parameters**
This callable has no explicit inputs.
**Returns**
- Type: `str`
- Direct return expressions: `result.stdout.strip()`; `'Apple Silicon'`
**Exceptions and behavior**
Function `_get_apple_chip_name` calls `subprocess.run`, `result.stdout.strip`; has 2 explicit return paths.
No direct `raise` statement appears in this definition.
[View source #L24-L35](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L24-L35).
vllm_mlx.vllm_platform._get_unified_memory_size · function
```python
vllm_mlx.vllm_platform._get_unified_memory_size() -> int
```
Get the total unified memory size in bytes.
**Parameters**
This callable has no explicit inputs.
**Returns**
- Type: `int`
- Direct return expressions: `int(result.stdout.strip())`; `8 * 1024 * 1024 * 1024`
**Exceptions and behavior**
Function `_get_unified_memory_size` calls `subprocess.run`, `int`, `result.stdout.strip`; has 2 explicit return paths.
No direct `raise` statement appears in this definition.
[View source #L38-L50](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L38-L50).
vllm_mlx.vllm_platform._is_mlx_available · function
```python
vllm_mlx.vllm_platform._is_mlx_available() -> bool
```
Check if MLX is available and working.
**Parameters**
This callable has no explicit inputs.
**Returns**
- Type: `bool`
- Direct return expressions: `True`; `False`
**Exceptions and behavior**
Function `_is_mlx_available` calls `mx.array`, `logger.debug`; has 2 explicit return paths.
No direct `raise` statement appears in this definition.
[View source #L53-L63](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L53-L63).
vllm_mlx.vllm_platform._is_apple_silicon · function
```python
vllm_mlx.vllm_platform._is_apple_silicon() -> bool
```
Check if running on Apple Silicon.
**Parameters**
This callable has no explicit inputs.
**Returns**
- Type: `bool`
- Direct return expressions: `sys.platform == 'darwin' and platform.machine() == 'arm64'`
**Exceptions and behavior**
Function `_is_apple_silicon` calls `platform.machine`; returns `sys.platform == 'darwin' and platform.machine() == 'arm64'`.
No direct `raise` statement appears in this definition.
[View source #L66-L68](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L66-L68).
vllm_mlx.vllm_platform.MLXPlatform · class
```python
vllm_mlx.vllm_platform.MLXPlatform()
```
Platform implementation for Apple Silicon using MLX.
**Parameters**
This callable has no explicit inputs.
**Returns**
- Constructs: `vllm_mlx.vllm_platform.MLXPlatform`
**Exceptions and behavior**
Class `MLXPlatform` declares 30 direct member(s).
No direct `raise` statement appears in this definition.
[View source #L71-L351](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L71-L351).
vllm_mlx.vllm_platform.MLXPlatform._enum · method
```python
vllm_mlx.vllm_platform.MLXPlatform._enum() -> not annotated
```
Method `MLXPlatform._enum` returns `PlatformEnum.OOT`.
**Parameters**
This callable has no explicit inputs.
**Returns**
- Type: `not annotated`
- Direct return expressions: `PlatformEnum.OOT`
**Exceptions and behavior**
Method `MLXPlatform._enum` returns `PlatformEnum.OOT`.
No direct `raise` statement appears in this definition.
[View source #L90-L93](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L90-L93).
vllm_mlx.vllm_platform.MLXPlatform.supported_dtypes · method
```python
vllm_mlx.vllm_platform.MLXPlatform.supported_dtypes() -> list[torch.dtype]
```
Return supported dtypes for MLX.
**Parameters**
This callable has no explicit inputs.
**Returns**
- Type: `list[torch.dtype]`
- Direct return expressions: `[torch.bfloat16, torch.float16, torch.float32]`; `[torch.float16, torch.float32]`
**Exceptions and behavior**
Method `MLXPlatform.supported_dtypes` calls `mx.array`; has 2 explicit return paths.
No direct `raise` statement appears in this definition.
[View source #L122-L133](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L122-L133).
vllm_mlx.vllm_platform.MLXPlatform.is_cuda · method
```python
vllm_mlx.vllm_platform.MLXPlatform.is_cuda() -> bool
```
Return ``False`` because this platform does not use CUDA.
**Parameters**
This callable has no explicit inputs.
**Returns**
- Type: `bool`
- Direct return expressions: `False`
**Exceptions and behavior**
Method `MLXPlatform.is_cuda` returns `False`.
No direct `raise` statement appears in this definition.
[View source #L135-L138](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L135-L138).
vllm_mlx.vllm_platform.MLXPlatform.is_rocm · method
```python
vllm_mlx.vllm_platform.MLXPlatform.is_rocm() -> bool
```
Return ``False`` because this platform does not use ROCm.
**Parameters**
This callable has no explicit inputs.
**Returns**
- Type: `bool`
- Direct return expressions: `False`
**Exceptions and behavior**
Method `MLXPlatform.is_rocm` returns `False`.
No direct `raise` statement appears in this definition.
[View source #L140-L143](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L140-L143).
vllm_mlx.vllm_platform.MLXPlatform.is_tpu · method
```python
vllm_mlx.vllm_platform.MLXPlatform.is_tpu() -> bool
```
Return ``False`` because this platform is not a TPU backend.
**Parameters**
This callable has no explicit inputs.
**Returns**
- Type: `bool`
- Direct return expressions: `False`
**Exceptions and behavior**
Method `MLXPlatform.is_tpu` returns `False`.
No direct `raise` statement appears in this definition.
[View source #L145-L148](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L145-L148).
vllm_mlx.vllm_platform.MLXPlatform.is_xpu · method
```python
vllm_mlx.vllm_platform.MLXPlatform.is_xpu() -> bool
```
Return ``False`` because this platform does not use Intel XPU.
**Parameters**
This callable has no explicit inputs.
**Returns**
- Type: `bool`
- Direct return expressions: `False`
**Exceptions and behavior**
Method `MLXPlatform.is_xpu` returns `False`.
No direct `raise` statement appears in this definition.
[View source #L150-L153](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L150-L153).
vllm_mlx.vllm_platform.MLXPlatform.is_cpu · method
```python
vllm_mlx.vllm_platform.MLXPlatform.is_cpu() -> bool
```
Return ``False`` because MLX targets Apple GPU acceleration here.
**Parameters**
This callable has no explicit inputs.
**Returns**
- Type: `bool`
- Direct return expressions: `False`
**Exceptions and behavior**
Method `MLXPlatform.is_cpu` returns `False`.
No direct `raise` statement appears in this definition.
[View source #L155-L158](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L155-L158).
vllm_mlx.vllm_platform.MLXPlatform.is_mlx · method
```python
vllm_mlx.vllm_platform.MLXPlatform.is_mlx() -> bool
```
Return ``True`` to identify the MLX platform plugin.
**Parameters**
This callable has no explicit inputs.
**Returns**
- Type: `bool`
- Direct return expressions: `True`
**Exceptions and behavior**
Method `MLXPlatform.is_mlx` returns `True`.
No direct `raise` statement appears in this definition.
[View source #L160-L163](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L160-L163).
vllm_mlx.vllm_platform.MLXPlatform.is_out_of_tree · method
```python
vllm_mlx.vllm_platform.MLXPlatform.is_out_of_tree() -> bool
```
Return ``True`` because MLX is registered as a vLLM plugin.
**Parameters**
This callable has no explicit inputs.
**Returns**
- Type: `bool`
- Direct return expressions: `True`
**Exceptions and behavior**
Method `MLXPlatform.is_out_of_tree` returns `True`.
No direct `raise` statement appears in this definition.
[View source #L165-L168](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L165-L168).
vllm_mlx.vllm_platform.MLXPlatform.is_cuda_alike · method
```python
vllm_mlx.vllm_platform.MLXPlatform.is_cuda_alike() -> bool
```
Return ``False`` because MLX does not implement CUDA semantics.
**Parameters**
This callable has no explicit inputs.
**Returns**
- Type: `bool`
- Direct return expressions: `False`
**Exceptions and behavior**
Method `MLXPlatform.is_cuda_alike` returns `False`.
No direct `raise` statement appears in this definition.
[View source #L170-L173](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L170-L173).
vllm_mlx.vllm_platform.MLXPlatform.is_sleep_mode_available · method
```python
vllm_mlx.vllm_platform.MLXPlatform.is_sleep_mode_available() -> bool
```
Return ``False`` because vLLM sleep mode is unavailable on MLX.
**Parameters**
This callable has no explicit inputs.
**Returns**
- Type: `bool`
- Direct return expressions: `False`
**Exceptions and behavior**
Method `MLXPlatform.is_sleep_mode_available` returns `False`.
No direct `raise` statement appears in this definition.
[View source #L175-L178](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L175-L178).
vllm_mlx.vllm_platform.MLXPlatform.get_device_name · method
```python
vllm_mlx.vllm_platform.MLXPlatform.get_device_name(device_id: int = 0) -> str
```
Get the Apple Silicon chip name.
**Parameters**
| Name | Type | Required | Default | Description |
| --- | --- | --- | --- | --- |
| `device_id` | `int` | `no` | `0` | Optional positional or keyword input; defaults to `0`. |
**Returns**
- Type: `str`
- Direct return expressions: `_get_apple_chip_name()`
**Exceptions and behavior**
Method `MLXPlatform.get_device_name` calls `_get_apple_chip_name`; returns `_get_apple_chip_name()`.
No direct `raise` statement appears in this definition.
[View source #L181-L183](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L181-L183).
vllm_mlx.vllm_platform.MLXPlatform.get_device_uuid · method
```python
vllm_mlx.vllm_platform.MLXPlatform.get_device_uuid(device_id: int = 0) -> str
```
Get device UUID (not applicable for MLX).
**Parameters**
| Name | Type | Required | Default | Description |
| --- | --- | --- | --- | --- |
| `device_id` | `int` | `no` | `0` | Optional positional or keyword input; defaults to `0`. |
**Returns**
- Type: `str`
- Direct return expressions: `'mlx-0'`
**Exceptions and behavior**
Method `MLXPlatform.get_device_uuid` returns `'mlx-0'`.
No direct `raise` statement appears in this definition.
[View source #L186-L188](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L186-L188).
vllm_mlx.vllm_platform.MLXPlatform.get_device_total_memory · method
```python
vllm_mlx.vllm_platform.MLXPlatform.get_device_total_memory(device_id: int = 0) -> int
```
Get total unified memory in bytes.
**Parameters**
| Name | Type | Required | Default | Description |
| --- | --- | --- | --- | --- |
| `device_id` | `int` | `no` | `0` | Optional positional or keyword input; defaults to `0`. |
**Returns**
- Type: `int`
- Direct return expressions: `_get_unified_memory_size()`
**Exceptions and behavior**
Method `MLXPlatform.get_device_total_memory` calls `_get_unified_memory_size`; returns `_get_unified_memory_size()`.
No direct `raise` statement appears in this definition.
[View source #L191-L193](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L191-L193).
vllm_mlx.vllm_platform.MLXPlatform.inference_mode · method
```python
vllm_mlx.vllm_platform.MLXPlatform.inference_mode() -> not annotated
```
Return inference mode context manager.
**Parameters**
This callable has no explicit inputs.
**Returns**
- Type: `not annotated`
- Direct return expressions: `torch.no_grad()`
**Exceptions and behavior**
Method `MLXPlatform.inference_mode` calls `torch.no_grad`; returns `torch.no_grad()`.
No direct `raise` statement appears in this definition.
[View source #L196-L200](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L196-L200).
vllm_mlx.vllm_platform.MLXPlatform.set_device · method
```python
vllm_mlx.vllm_platform.MLXPlatform.set_device(device: torch.device) -> None
```
Set the device (no-op for MLX, uses default device).
**Parameters**
| Name | Type | Required | Default | Description |
| --- | --- | --- | --- | --- |
| `device` | `torch.device` | `yes` | `none` | Required positional or keyword input. |
**Returns**
- Type: `None`
**Exceptions and behavior**
Method `MLXPlatform.set_device` contains no state mutation, call, raise, return, await, or yield.
No direct `raise` statement appears in this definition.
[View source #L203-L206](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L203-L206).
vllm_mlx.vllm_platform.MLXPlatform.seed_everything · method
```python
vllm_mlx.vllm_platform.MLXPlatform.seed_everything(seed: int | None = None) -> None
```
Set random seeds for reproducibility.
**Parameters**
| Name | Type | Required | Default | Description |
| --- | --- | --- | --- | --- |
| `seed` | `int \| None` | `no` | `None` | Optional positional or keyword input; defaults to `None`. |
**Returns**
- Type: `None`
**Exceptions and behavior**
Method `MLXPlatform.seed_everything` calls `random.seed`, `np.random.seed`, `torch.manual_seed`, `mx.random.seed`.
No direct `raise` statement appears in this definition.
[View source #L209-L225](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L209-L225).
vllm_mlx.vllm_platform.MLXPlatform.import_kernels · method
```python
vllm_mlx.vllm_platform.MLXPlatform.import_kernels() -> None
```
Import MLX kernels (no custom C kernels).
**Parameters**
This callable has no explicit inputs.
**Returns**
- Type: `None`
**Exceptions and behavior**
Method `MLXPlatform.import_kernels` contains no state mutation, call, raise, return, await, or yield.
No direct `raise` statement appears in this definition.
[View source #L228-L231](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L228-L231).
vllm_mlx.vllm_platform.MLXPlatform.get_attn_backend_cls · method
```python
vllm_mlx.vllm_platform.MLXPlatform.get_attn_backend_cls(selected_backend, head_size: int, dtype: torch.dtype, kv_cache_dtype, block_size: int, use_mla: bool, has_sink: bool, use_sparse: bool, attn_type: str | None = None) -> str
```
Return MLX attention backend class path.
**Parameters**
| Name | Type | Required | Default | Description |
| --- | --- | --- | --- | --- |
| `selected_backend` | `not annotated` | `yes` | `none` | Required positional or keyword input. |
| `head_size` | `int` | `yes` | `none` | Required positional or keyword input. |
| `dtype` | `torch.dtype` | `yes` | `none` | Required positional or keyword input. |
| `kv_cache_dtype` | `not annotated` | `yes` | `none` | Required positional or keyword input. |
| `block_size` | `int` | `yes` | `none` | Required positional or keyword input. |
| `use_mla` | `bool` | `yes` | `none` | Required positional or keyword input. |
| `has_sink` | `bool` | `yes` | `none` | Required positional or keyword input. |
| `use_sparse` | `bool` | `yes` | `none` | Required positional or keyword input. |
| `attn_type` | `str \| None` | `no` | `None` | Optional positional or keyword input; defaults to `None`. |
**Returns**
- Type: `str`
- Direct return expressions: `'vllm_mlx.attention.MLXAttentionBackend'`
**Exceptions and behavior**
Method `MLXPlatform.get_attn_backend_cls` returns `'vllm_mlx.attention.MLXAttentionBackend'`.
No direct `raise` statement appears in this definition.
[View source #L234-L248](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L234-L248).
vllm_mlx.vllm_platform.MLXPlatform.check_and_update_config · method
```python
vllm_mlx.vllm_platform.MLXPlatform.check_and_update_config(vllm_config: 'VllmConfig') -> None
```
Check and update vLLM configuration for MLX.
**Parameters**
| Name | Type | Required | Default | Description |
| --- | --- | --- | --- | --- |
| `vllm_config` | `'VllmConfig'` | `yes` | `none` | Required positional or keyword input. |
**Returns**
- Type: `None`
**Exceptions and behavior**
Method `MLXPlatform.check_and_update_config` calls `logger.info`, `_get_apple_chip_name`, `_get_unified_memory_size`, `hasattr`.
No direct `raise` statement appears in this definition.
[View source #L251-L280](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L251-L280).
vllm_mlx.vllm_platform.MLXPlatform.verify_model_arch · method
```python
vllm_mlx.vllm_platform.MLXPlatform.verify_model_arch(model_arch: str) -> None
```
Verify model architecture is supported on MLX.
**Parameters**
| Name | Type | Required | Default | Description |
| --- | --- | --- | --- | --- |
| `model_arch` | `str` | `yes` | `none` | Required positional or keyword input. |
**Returns**
- Type: `None`
**Exceptions and behavior**
Method `MLXPlatform.verify_model_arch` calls `hint.lower`, `model_arch.lower`, `logger.warning`.
No direct `raise` statement appears in this definition.
[View source #L283-L294](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L283-L294).
vllm_mlx.vllm_platform.MLXPlatform.verify_quantization · method
```python
vllm_mlx.vllm_platform.MLXPlatform.verify_quantization(quant: str) -> None
```
Verify quantization method is supported.
**Parameters**
| Name | Type | Required | Default | Description |
| --- | --- | --- | --- | --- |
| `quant` | `str` | `yes` | `none` | Required positional or keyword input. |
**Returns**
- Type: `None`
**Exceptions and behavior**
Method `MLXPlatform.verify_quantization` calls `ValueError`; can raise `ValueError`.
Directly raised exceptions: `ValueError`.
[View source #L297-L304](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L297-L304).
vllm_mlx.vllm_platform.MLXPlatform.is_pin_memory_available · method
```python
vllm_mlx.vllm_platform.MLXPlatform.is_pin_memory_available() -> bool
```
Pin memory not needed with unified memory.
**Parameters**
This callable has no explicit inputs.
**Returns**
- Type: `bool`
- Direct return expressions: `False`
**Exceptions and behavior**
Method `MLXPlatform.is_pin_memory_available` returns `False`.
No direct `raise` statement appears in this definition.
[View source #L307-L309](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L307-L309).
vllm_mlx.vllm_platform.MLXPlatform.get_current_memory_usage · method
```python
vllm_mlx.vllm_platform.MLXPlatform.get_current_memory_usage(device = None) -> float
```
Get current memory usage in bytes.
**Parameters**
| Name | Type | Required | Default | Description |
| --- | --- | --- | --- | --- |
| `device` | `not annotated` | `no` | `None` | Optional positional or keyword input; defaults to `None`. |
**Returns**
- Type: `float`
- Direct return expressions: `float(process.memory_info().rss)`; `0.0`
**Exceptions and behavior**
Method `MLXPlatform.get_current_memory_usage` calls `psutil.Process`, `float`, `process.memory_info`; has 2 explicit return paths.
No direct `raise` statement appears in this definition.
[View source #L312-L323](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L312-L323).
vllm_mlx.vllm_platform.MLXPlatform.supports_fp8 · method
```python
vllm_mlx.vllm_platform.MLXPlatform.supports_fp8() -> bool
```
FP8 not supported on MLX.
**Parameters**
This callable has no explicit inputs.
**Returns**
- Type: `bool`
- Direct return expressions: `False`
**Exceptions and behavior**
Method `MLXPlatform.supports_fp8` returns `False`.
No direct `raise` statement appears in this definition.
[View source #L326-L328](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L326-L328).
vllm_mlx.vllm_platform.MLXPlatform.use_custom_allreduce · method
```python
vllm_mlx.vllm_platform.MLXPlatform.use_custom_allreduce() -> bool
```
Custom allreduce not available.
**Parameters**
This callable has no explicit inputs.
**Returns**
- Type: `bool`
- Direct return expressions: `False`
**Exceptions and behavior**
Method `MLXPlatform.use_custom_allreduce` returns `False`.
No direct `raise` statement appears in this definition.
[View source #L331-L333](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L331-L333).
vllm_mlx.vllm_platform.MLXPlatform.support_static_graph_mode · method
```python
vllm_mlx.vllm_platform.MLXPlatform.support_static_graph_mode() -> bool
```
Static graph mode (CUDA graphs) not supported.
**Parameters**
This callable has no explicit inputs.
**Returns**
- Type: `bool`
- Direct return expressions: `False`
**Exceptions and behavior**
Method `MLXPlatform.support_static_graph_mode` returns `False`.
No direct `raise` statement appears in this definition.
[View source #L336-L338](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L336-L338).
vllm_mlx.vllm_platform.MLXPlatform.get_device_communicator_cls · method
```python
vllm_mlx.vllm_platform.MLXPlatform.get_device_communicator_cls() -> str
```
Return the communicator class for distributed.
**Parameters**
This callable has no explicit inputs.
**Returns**
- Type: `str`
- Direct return expressions: `'vllm_mlx.distributed.MLXCommunicator'`
**Exceptions and behavior**
Method `MLXPlatform.get_device_communicator_cls` returns `'vllm_mlx.distributed.MLXCommunicator'`.
No direct `raise` statement appears in this definition.
[View source #L341-L343](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L341-L343).
vllm_mlx.vllm_platform.MLXPlatform.get_punica_wrapper · method
```python
vllm_mlx.vllm_platform.MLXPlatform.get_punica_wrapper() -> str
```
Return LoRA wrapper (not yet implemented for MLX).
**Parameters**
This callable has no explicit inputs.
**Returns**
- Type: `str`
**Exceptions and behavior**
Method `MLXPlatform.get_punica_wrapper` calls `NotImplementedError`; can raise `NotImplementedError`.
Directly raised exceptions: `NotImplementedError`.
[View source #L346-L348](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L346-L348).
vllm_mlx.vllm_platform.MLXPlatform.__repr__ · method
```python
vllm_mlx.vllm_platform.MLXPlatform.__repr__() -> str
```
Method `MLXPlatform.__repr__` returns `f''`.
**Parameters**
This callable has no explicit inputs.
**Returns**
- Type: `str`
- Direct return expressions: `f''`
**Exceptions and behavior**
Method `MLXPlatform.__repr__` returns `f''`.
No direct `raise` statement appears in this definition.
[View source #L350-L351](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L350-L351).
## Complete symbol map
This map also includes private definitions and nested helpers. The signature column exposes every explicit input even when an internal helper has no dedicated parameter prose.
| Symbol | Kind | Signature and inputs | What it does | Source |
| --- | --- | --- | --- | --- |
| [`_get_apple_chip_name`](#contract-vllm_mlx.vllm_platform._get_apple_chip_name) | function | `_get_apple_chip_name() -> str` | Get the name of the Apple Silicon chip. | [#L24-L35](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L24-L35) |
| [`_get_unified_memory_size`](#contract-vllm_mlx.vllm_platform._get_unified_memory_size) | function | `_get_unified_memory_size() -> int` | Get the total unified memory size in bytes. | [#L38-L50](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L38-L50) |
| [`_is_mlx_available`](#contract-vllm_mlx.vllm_platform._is_mlx_available) | function | `_is_mlx_available() -> bool` | Check if MLX is available and working. | [#L53-L63](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L53-L63) |
| [`_is_apple_silicon`](#contract-vllm_mlx.vllm_platform._is_apple_silicon) | function | `_is_apple_silicon() -> bool` | Check if running on Apple Silicon. | [#L66-L68](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L66-L68) |
| [`MLXPlatform`](#contract-vllm_mlx.vllm_platform.MLXPlatform) | class | `MLXPlatform()` | Platform implementation for Apple Silicon using MLX. | [#L71-L351](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L71-L351) |
| [`MLXPlatform._enum`](#contract-vllm_mlx.vllm_platform.MLXPlatform._enum) | method | `MLXPlatform._enum() -> not annotated` | Method `MLXPlatform._enum` returns `PlatformEnum.OOT`. | [#L90-L93](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L90-L93) |
| [`MLXPlatform.supported_dtypes`](#contract-vllm_mlx.vllm_platform.MLXPlatform.supported_dtypes) | method | `MLXPlatform.supported_dtypes() -> list[torch.dtype]` | Return supported dtypes for MLX. | [#L122-L133](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L122-L133) |
| [`MLXPlatform.is_cuda`](#contract-vllm_mlx.vllm_platform.MLXPlatform.is_cuda) | method | `MLXPlatform.is_cuda() -> bool` | Return ``False`` because this platform does not use CUDA. | [#L135-L138](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L135-L138) |
| [`MLXPlatform.is_rocm`](#contract-vllm_mlx.vllm_platform.MLXPlatform.is_rocm) | method | `MLXPlatform.is_rocm() -> bool` | Return ``False`` because this platform does not use ROCm. | [#L140-L143](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L140-L143) |
| [`MLXPlatform.is_tpu`](#contract-vllm_mlx.vllm_platform.MLXPlatform.is_tpu) | method | `MLXPlatform.is_tpu() -> bool` | Return ``False`` because this platform is not a TPU backend. | [#L145-L148](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L145-L148) |
| [`MLXPlatform.is_xpu`](#contract-vllm_mlx.vllm_platform.MLXPlatform.is_xpu) | method | `MLXPlatform.is_xpu() -> bool` | Return ``False`` because this platform does not use Intel XPU. | [#L150-L153](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L150-L153) |
| [`MLXPlatform.is_cpu`](#contract-vllm_mlx.vllm_platform.MLXPlatform.is_cpu) | method | `MLXPlatform.is_cpu() -> bool` | Return ``False`` because MLX targets Apple GPU acceleration here. | [#L155-L158](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L155-L158) |
| [`MLXPlatform.is_mlx`](#contract-vllm_mlx.vllm_platform.MLXPlatform.is_mlx) | method | `MLXPlatform.is_mlx() -> bool` | Return ``True`` to identify the MLX platform plugin. | [#L160-L163](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L160-L163) |
| [`MLXPlatform.is_out_of_tree`](#contract-vllm_mlx.vllm_platform.MLXPlatform.is_out_of_tree) | method | `MLXPlatform.is_out_of_tree() -> bool` | Return ``True`` because MLX is registered as a vLLM plugin. | [#L165-L168](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L165-L168) |
| [`MLXPlatform.is_cuda_alike`](#contract-vllm_mlx.vllm_platform.MLXPlatform.is_cuda_alike) | method | `MLXPlatform.is_cuda_alike() -> bool` | Return ``False`` because MLX does not implement CUDA semantics. | [#L170-L173](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L170-L173) |
| [`MLXPlatform.is_sleep_mode_available`](#contract-vllm_mlx.vllm_platform.MLXPlatform.is_sleep_mode_available) | method | `MLXPlatform.is_sleep_mode_available() -> bool` | Return ``False`` because vLLM sleep mode is unavailable on MLX. | [#L175-L178](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L175-L178) |
| [`MLXPlatform.get_device_name`](#contract-vllm_mlx.vllm_platform.MLXPlatform.get_device_name) | method | `MLXPlatform.get_device_name(device_id: int = 0) -> str` | Get the Apple Silicon chip name. | [#L181-L183](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L181-L183) |
| [`MLXPlatform.get_device_uuid`](#contract-vllm_mlx.vllm_platform.MLXPlatform.get_device_uuid) | method | `MLXPlatform.get_device_uuid(device_id: int = 0) -> str` | Get device UUID (not applicable for MLX). | [#L186-L188](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L186-L188) |
| [`MLXPlatform.get_device_total_memory`](#contract-vllm_mlx.vllm_platform.MLXPlatform.get_device_total_memory) | method | `MLXPlatform.get_device_total_memory(device_id: int = 0) -> int` | Get total unified memory in bytes. | [#L191-L193](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L191-L193) |
| [`MLXPlatform.inference_mode`](#contract-vllm_mlx.vllm_platform.MLXPlatform.inference_mode) | method | `MLXPlatform.inference_mode() -> not annotated` | Return inference mode context manager. | [#L196-L200](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L196-L200) |
| [`MLXPlatform.set_device`](#contract-vllm_mlx.vllm_platform.MLXPlatform.set_device) | method | `MLXPlatform.set_device(device: torch.device) -> None` | Set the device (no-op for MLX, uses default device). | [#L203-L206](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L203-L206) |
| [`MLXPlatform.seed_everything`](#contract-vllm_mlx.vllm_platform.MLXPlatform.seed_everything) | method | `MLXPlatform.seed_everything(seed: int \| None = None) -> None` | Set random seeds for reproducibility. | [#L209-L225](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L209-L225) |
| [`MLXPlatform.import_kernels`](#contract-vllm_mlx.vllm_platform.MLXPlatform.import_kernels) | method | `MLXPlatform.import_kernels() -> None` | Import MLX kernels (no custom C kernels). | [#L228-L231](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L228-L231) |
| [`MLXPlatform.get_attn_backend_cls`](#contract-vllm_mlx.vllm_platform.MLXPlatform.get_attn_backend_cls) | method | `MLXPlatform.get_attn_backend_cls(selected_backend, head_size: int, dtype: torch.dtype, kv_cache_dtype, block_size: int, use_mla: bool, has_sink: bool, use_sparse: bool, attn_type: str \| None = None) -> str` | Return MLX attention backend class path. | [#L234-L248](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L234-L248) |
| [`MLXPlatform.check_and_update_config`](#contract-vllm_mlx.vllm_platform.MLXPlatform.check_and_update_config) | method | `MLXPlatform.check_and_update_config(vllm_config: 'VllmConfig') -> None` | Check and update vLLM configuration for MLX. | [#L251-L280](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L251-L280) |
| [`MLXPlatform.verify_model_arch`](#contract-vllm_mlx.vllm_platform.MLXPlatform.verify_model_arch) | method | `MLXPlatform.verify_model_arch(model_arch: str) -> None` | Verify model architecture is supported on MLX. | [#L283-L294](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L283-L294) |
| [`MLXPlatform.verify_quantization`](#contract-vllm_mlx.vllm_platform.MLXPlatform.verify_quantization) | method | `MLXPlatform.verify_quantization(quant: str) -> None` | Verify quantization method is supported. | [#L297-L304](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L297-L304) |
| [`MLXPlatform.is_pin_memory_available`](#contract-vllm_mlx.vllm_platform.MLXPlatform.is_pin_memory_available) | method | `MLXPlatform.is_pin_memory_available() -> bool` | Pin memory not needed with unified memory. | [#L307-L309](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L307-L309) |
| [`MLXPlatform.get_current_memory_usage`](#contract-vllm_mlx.vllm_platform.MLXPlatform.get_current_memory_usage) | method | `MLXPlatform.get_current_memory_usage(device = None) -> float` | Get current memory usage in bytes. | [#L312-L323](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L312-L323) |
| [`MLXPlatform.supports_fp8`](#contract-vllm_mlx.vllm_platform.MLXPlatform.supports_fp8) | method | `MLXPlatform.supports_fp8() -> bool` | FP8 not supported on MLX. | [#L326-L328](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L326-L328) |
| [`MLXPlatform.use_custom_allreduce`](#contract-vllm_mlx.vllm_platform.MLXPlatform.use_custom_allreduce) | method | `MLXPlatform.use_custom_allreduce() -> bool` | Custom allreduce not available. | [#L331-L333](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L331-L333) |
| [`MLXPlatform.support_static_graph_mode`](#contract-vllm_mlx.vllm_platform.MLXPlatform.support_static_graph_mode) | method | `MLXPlatform.support_static_graph_mode() -> bool` | Static graph mode (CUDA graphs) not supported. | [#L336-L338](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L336-L338) |
| [`MLXPlatform.get_device_communicator_cls`](#contract-vllm_mlx.vllm_platform.MLXPlatform.get_device_communicator_cls) | method | `MLXPlatform.get_device_communicator_cls() -> str` | Return the communicator class for distributed. | [#L341-L343](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L341-L343) |
| [`MLXPlatform.get_punica_wrapper`](#contract-vllm_mlx.vllm_platform.MLXPlatform.get_punica_wrapper) | method | `MLXPlatform.get_punica_wrapper() -> str` | Return LoRA wrapper (not yet implemented for MLX). | [#L346-L348](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L346-L348) |
| [`MLXPlatform.__repr__`](#contract-vllm_mlx.vllm_platform.MLXPlatform.__repr__) | method | `MLXPlatform.__repr__() -> str` | Method `MLXPlatform.__repr__` returns `f''`. | [#L350-L351](https://github.com/waybarrios/vllm-mlx/blob/a69d47912bcb21d8fe04d48f75fa896b620ffcfa/vllm_mlx/vllm_platform.py#L350-L351) |