经 AI Skill Hub 精选评估,苹果芯片LLM运行时 获评「强烈推荐」。这款AI工具在功能完整性、社区活跃度和易用性方面表现出色,AI 评分 8.0 分,适合有一定技术背景的用户使用。
支持Gemma 4和Qwen 3.6 MTP模式的开源AI工具
苹果芯片LLM运行时 是一款基于 Rust 开发的开源工具,专注于 ai-interface、gemma4、generative-ai 等核心功能。作为 GitHub 开源项目,它拥有活跃的社区支持和持续的版本迭代,代码完全透明可审计,支持本地部署以保护数据隐私。无论是个人使用还是集成到企业工作流,都能提供稳定可靠的解决方案。
支持Gemma 4和Qwen 3.6 MTP模式的开源AI工具
苹果芯片LLM运行时 是一款基于 Rust 开发的开源工具,专注于 ai-interface、gemma4、generative-ai 等核心功能。作为 GitHub 开源项目,它拥有活跃的社区支持和持续的版本迭代,代码完全透明可审计,支持本地部署以保护数据隐私。无论是个人使用还是集成到企业工作流,都能提供稳定可靠的解决方案。
# 方式一:cargo install(推荐) cargo install ax-engine # 方式二:从源码编译 git clone https://github.com/defai-digital/ax-engine cd ax-engine cargo build --release # 二进制在 ./target/release/ax-engine
# 查看帮助 ax-engine --help # 基本运行 ax-engine [options] <input> # 详细使用说明请查阅文档 # https://github.com/defai-digital/ax-engine
# ax-engine 配置说明 # 查看配置选项 ax-engine --config-example > config.yml # 常见配置项 # output_dir: ./output # log_level: info # workers: 4 # 环境变量(覆盖配置文件) export AX_ENGINE_CONFIG="/path/to/config.yml"
AX Engine is a Mac-first LLM inference runtime for Apple Silicon. Install with Homebrew, download a curated model, and serve OpenAI-compatible endpoints locally — with a repo-owned MLX path for Gemma, Qwen, and GLM, first-class MTP, multi-model serving with exact-prompt prefix reuse, and peer-backed benchmarks against mlx-lm, MTPLX, and OMLX.
NVIDIA/CUDA fleet serving lives in AX Serving. AX Engine remains the local Apple Silicon runtime and no longer ships the former vLLM or TensorRT provider bridges, runtime package, container, or CUDA qualification scripts.
Browse AutomatosX serve-ready chat / coding / embedding snapshots in the AutomatosX model collection on Hugging Face. Additional native families (GLM 4.7 Flash, Nemotron Omni, Unlimited-OCR, Whisper, MiniCPM-V, and others) are documented under Supported Models.
Requires macOS 26 (Tahoe)+ on Apple Silicon (M2 or newer). For compact single models (Qwen 3.5 9B 4-bit preferred; 6-bit also fits), 16 GB unified memory is enough — including base Mac mini M4 16 GB. Prefer 4-bit for headroom. For multi-model serving, longer contexts, and larger packs (27B/35B class), plan on 32 GB+ (64 GB recommended).
Use the wheel for Python applications that import ax_engine, optional Python integrations, or systems where Homebrew is unavailable. Install it in a virtual environment:
python3 -m venv .venv
source .venv/bin/activate
python3 -m pip install --upgrade pip
python3 -m pip install --upgrade "ax-engine[download]>=7.3.0,<8"
ax-engine doctor
The wheel also exposes ax-engine and ax-engine-server and bundles the bench binary used by diagnostics. If both Homebrew and pip are installed, an active virtual environment normally wins on PATH; use which -a ax-engine to see every copy and prefer one installation channel in each shell. See Getting Started for the full channel comparison and troubleshooting.
Most clients use the OpenAI-compatible HTTP server. Python also has an in-process session API.
| SDK | Docs |
|---|---|
| Rust | [docs/sdk/rust.md](docs/sdk/rust.md) |
| Python | [docs/sdk/python.md](docs/sdk/python.md) |
| JavaScript / TypeScript | [docs/sdk/javascript.md](docs/sdk/javascript.md) |
| Go | [docs/sdk/go.md](docs/sdk/go.md) |
| Ruby | [docs/sdk/ruby.md](docs/sdk/ruby.md) |
| Swift | [docs/sdk/swift.md](docs/sdk/swift.md) |
| Mojo *(experimental)* | [docs/sdk/mojo.md](docs/sdk/mojo.md) |
Streaming OpenAI /v1/chat/completions — the comparison users run when they open a server and time chat. AX Engine 6.13.1 vs peer MLX serving engine 0.4.3, Apple M5 Max 128 GB, Qwen 3.6 27B / 35B-A3B at 4-bit and 6-bit, ~512 and ~2k prompt targets, 256 gen tokens, temperature 0.
| Model | p512 decode (AX / peer) | p2048 decode (AX / peer) |
|---|---|---|
| Qwen3.6 27B 4-bit | **34.40 / 32.32 (+6.4%)** | **33.88 / 32.01 (+5.9%)** |
| Qwen3.6 27B 6-bit | **24.59 / 23.94 (+2.7%)** | **23.97 / 23.35 (+2.7%)** |
| Qwen3.6 35B-A3B 4-bit | **159.10 / 129.06 (+23.3%)** | **156.89 / 126.60 (+23.9%)** |
| Qwen3.6 35B-A3B 6-bit | **128.79 / 106.67 (+20.7%)** | **126.90 / 105.04 (+20.8%)** |
AX wins 8 of 8 decode cells; geometric-mean decode advantage is 12.9% (dense 27B 4.4%, 35B-A3B MoE 22.2%). Effective prefill and TTFT split 4/8 and are roughly neutral in the matrix-wide geometric mean, so they are not headline wins. Full prefill/TTFT tables, methodology, provenance, and caveats:
This is the current AXQ campaign on df-macbookpro-m5 (Apple M5 Max, 128 GB, macOS 26.6.2). It uses the repository flappy prompt suite, four prompt cases, 256 generated tokens, greedy sampling, two warmups, five measured repetitions, three-second cooldowns, and disabled prefix-cache/n-gram stacking. Values are the median decode throughput over 20 measured runs.
The requested Qwen3.6 25B and Gemma4 35B labels do not correspond to published AutomatosX AXQ packs. The measured pack mappings are Qwen3.6 27B and Gemma4 31B, respectively. Exact raw artifacts and runtime caveats are in the campaign result.
<img width="100%" src="docs/assets/perf-mtp-peer-comparison-apples-to-apples.svg" alt="AXQ MTP decode throughput on Apple M5 Max comparing AX Engine, MTPLX, and OMLX">
| AXQ model | AX Engine 7.2.0 | MTPLX 2.9.0 | OMLX 0.6.4 | Readout |
|---|---|---|---|---|
| Qwen3.8 27B 6-bit | **45.05 tok/s** | 46.68 tok/s | 37.04 tok/s | AX exact MTP; MTPLX accepted 100% of drafted tokens; OMLX text-only staging |
| Qwen3.6 27B 6-bit *(requested 25B)* | **45.53 tok/s** | 46.86 tok/s | 38.39 tok/s | AX exact MTP; MTPLX accepted 99.51% of drafted tokens; OMLX text-only staging |
| Gemma4 31B 6-bit *(requested 35B)* | **22.49 tok/s** | unsupported | unsupported | AX assistant-MTP depth 2; peers rejected the AXQ vision/assistant contract |
| Gemma4 26B-A4B 6-bit | **112.70 tok/s** | unsupported | unsupported | AX assistant-MTP depth 2; one AX telemetry row was incomplete |
The OMLX Qwen rows use BatchedEngine with mtp_enabled and an imported AXQ MTP sidecar; its VLM loader rejected this AXQ vision-key layout, so these are text-only OMLX measurements. MTPLX rejected both Gemma packs because they declare an MTP layer but do not ship MTPLX-compatible root MTP weights. No unsupported lane is replaced with direct-mode throughput.
Per-runtime raw artifacts and the full contract: AXQ MTP peer campaign.
AX Engine 是一款专为 Apple Silicon 设计的本地 LLM 推理运行时、本地服务器、SDK 层和基准工具集。它可以直接支持 MLX 模型家族,并将其他 MLX 文本模型或非 MLX 模型通过显式 mlx-lm 和 llama.cpp 兼容性路由进行路由。
AX Engine 的主要功能包括:提供本地 OpenAI 兼容的模型服务器,支持常见的聊天和完成流程,提供 SDKs 支持 Python、TypeScript/JavaScript、Go、Ruby 和 Mojo 等语言。AX Engine 还支持直接支持的 Gemma 和 Qwen 模型家族,以及显式路由的非直接支持模型。
AX Engine 没有明确的环境依赖或系统要求说明。
安装 AX Engine 可以使用以下步骤:使用 pip 安装 ax-engine[download] 包,下载一个小型模型并启动服务器,使用高内存模型快捷方式等。
使用 AX Engine 的快速入门步骤包括:安装 ax-engine[download] 包,下载一个小型模型并启动服务器,使用高内存模型快捷方式等。
AX Engine 支持将下载的模型复制到一个显式目录中,使用 MCP 复制模型到一个显式目录中。
AX Engine 提供 Python SDK,支持下载模型,使用 built-in 下载别名等功能。
AX Engine 支持的模型包括直接支持的 Gemma 和 Qwen 模型家族,以及显式路由的非直接支持模型。
高性能AI模型运行时,支持多种模式
AI Skill Hub 为第三方内容聚合平台,本页面信息基于公开数据整理,不对工具功能和质量作任何法律背书。
建议在沙箱或测试环境中充分验证后,再部署至生产环境,并做好必要的安全评估。
✅ Apache 2.0 — 宽松开源协议,可商用,需保留版权声明和 NOTICE 文件,含专利授权条款。
AI Skill Hub 点评:苹果芯片LLM运行时 的核心功能完整,质量优秀。对于AI 技术爱好者来说,这是一个值得纳入个人工具库的选择。建议先在非生产环境试用,再逐步推广。
| 原始名称 | ax-engine |
| 原始描述 | 开源AI工具:Apple Silicon LLM runtime supporting Gemma 4 and Qwen 3.6 MTP modes。⭐9 · Rust |
| Topics | ai-interfacegemma4generative-aiinference-enginellmrust |
| GitHub | https://github.com/defai-digital/ax-engine |
| License | Apache-2.0 |
| 语言 | Rust |
收录时间:2026-06-12 · 更新时间:2026-06-13 · License:Apache-2.0 · AI Skill Hub 不对第三方内容的准确性作法律背书。