mlx-examples/llms/mlx_lm/SERVER.md

HTTP Model Server

You use mlx-lm to make an HTTP API for generating text with any supported model. The HTTP API is intended to be similar to the OpenAI chat API.

Note

The MLX LM server is not recommended for production as it only implements basic security checks.

Start the server with:

mlx_lm.server --model <path_to_model_or_hf_repo>

For example:

mlx_lm.server --model mlx-community/Mistral-7B-Instruct-v0.3-4bit

This will start a text generation server on port 8080 of the localhost using Mistral 7B instruct. The model will be downloaded from the provided Hugging Face repo if it is not already in the local cache.

To see a full list of options run:

mlx_lm.server --help

You can make a request to the model by running:

curl localhost:8080/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
     "messages": [{"role": "user", "content": "Say this is a test!"}],
     "temperature": 0.7
   }'

Request Fields

• messages: An array of message objects representing the conversation history. Each message object should have a role (e.g. user, assistant) and content (the message text).

• role_mapping: (Optional) A dictionary to customize the role prefixes in the generated prompt. If not provided, the default mappings are used.

• stop: (Optional) An array of strings or a single string. Thesse are sequences of tokens on which the generation should stop.

• max_tokens: (Optional) An integer specifying the maximum number of tokens to generate. Defaults to 100.

• stream: (Optional) A boolean indicating if the response should be streamed. If true, responses are sent as they are generated. Defaults to false.

• temperature: (Optional) A float specifying the sampling temperature. Defaults to 1.0.

• top_p: (Optional) A float specifying the nucleus sampling parameter. Defaults to 1.0.

• repetition_penalty: (Optional) Applies a penalty to repeated tokens. Defaults to 1.0.

• repetition_context_size: (Optional) The size of the context window for applying repetition penalty. Defaults to 20.

• logit_bias: (Optional) A dictionary mapping token IDs to their bias values. Defaults to None.

• logprobs: (Optional) An integer specifying the number of top tokens and corresponding log probabilities to return for each output in the generated sequence. If set, this can be any value between 1 and 10, inclusive.

• model: (Optional) A string path to a local model or Hugging Face repo id. If the path is local is must be relative to the directory the server was started in.

• adapters: (Optional) A string path to low-rank adapters. The path must be rlative to the directory the server was started in.

List Models

Use the v1/models endpoint to list available models:

curl localhost:8080/v1/models -H "Content-Type: application/json"

This will return a list of locally available models where each model in the list contains the following fields:

• "id": The Hugging Face repo id.
• "created": A timestamp representing the model creation time.