MLX

mlx-lm 是 Apple 的数组计算框架指向语言模型的形态。在 M 系列 Mac 上它是最快的本地路径，因为它用统一内存，不必和独立显卡抢资源。

仅限 Apple Silicon。`mlx_lm.server --model mlx-community/Qwen3-8B-4bit` 会在 localhost:8080/v1 上提供 OpenAI 兼容服务。上下文有多大，取决于你的统一内存。

最大上下文: Whatever your unified memory allows
免费额度: 无限且私密
注册要求: 什么都不用 —— 无账号

端点

接口地址http://localhost:8080/v1

速率限制None — it is your machine

OpenAI 兼容
官方 SDK：mlx-lm
文本
视觉

代码示例

npm install openai

import OpenAI from 'openai'

const client = new OpenAI({
  baseURL: 'http://localhost:8080/v1',
  apiKey: 'not-needed'
})

const response = await client.chat.completions.create({
  model: 'mlx-community/Qwen3-8B-4bit',
  messages: [{ role: 'user', content: 'Explain closures in one paragraph.' }]
})

console.log(response.choices[0].message.content)

pip install openai

from openai import OpenAI

client = OpenAI(
    base_url="http://localhost:8080/v1",
    api_key="not-needed",
)

response = client.chat.completions.create(
    model="mlx-community/Qwen3-8B-4bit",
    messages=[{"role": "user", "content": "Explain closures in one paragraph."}],
)

print(response.choices[0].message.content)

curl http://localhost:8080/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "mlx-community/Qwen3-8B-4bit",
    "messages": [{"role": "user", "content": "Explain closures in one paragraph."}]
  }'

pip install mlx-lm

from mlx_lm import load, generate

model, tokenizer = load("mlx-community/Qwen3-8B-4bit")

prompt = tokenizer.apply_chat_template(
    [{"role": "user", "content": "Explain closures in one paragraph."}],
    add_generation_prompt=True,
)

print(generate(model, tokenizer, prompt=prompt, max_tokens=512))

值得注意

仅限 Apple Silicon。
先起服务器：`mlx_lm.server --model mlx-community/Qwen3-8B-4bit`。

你能免费使用的模型3

Qwen 3qwen3
Qwen CoderQwen2.5-Coder-7B
GPT-OSS 120Bgpt-oss-120b