# Install rapid-mlx with Homebrew Fast local AI engine for Apple Silicon with an OpenAI-compatible API. Version 0.11.9 via Homebrew; verified 2026-08-03. ## Install ```sh sudo av install brew:rapid-mlx ``` Additional install commands: ### macOS - Homebrew (100%): ```sh brew install rapid-mlx ``` Evidence: local Homebrew formula metadata ## Package facts - **Package key:** brew:rapid-mlx - **Package manager:** Homebrew - **Version:** 0.11.9 - **Source summary:** Fast local AI engine for Apple Silicon with an OpenAI-compatible API - **Homepage:** - **Last updated:** 2026-08-03T09:11:37Z - **Generated:** 2026-08-03T19:37:03+00:00 ## Executables - rapid-mlx (alias) - rapid-mlx-bench (alias) - rapid-mlx-chat (alias) - rmlx (alias) - vllm-mlx (alias) - vllm-mlx-bench (alias) - vllm-mlx-chat (alias) ## Install behavior - Bottle: not available ## Freshness - Page generated: 2026-08-03 - Package-manager version: 0.11.9 ## Project history and usage Rapid-MLX is a local inference engine and OpenAI-compatible HTTP server built on Apple's MLX stack for Apple Silicon Macs. It packages model serving, interactive chat, benchmarking, model management, and agent/IDE integration behind the `rapid-mlx` CLI. ### Project history Development began in early 2026, with the official repository created in February and tagged releases available from March. The package was formerly exposed through `vllm-mlx` command names; those entry points remain as compatibility aliases while current documentation directs new users to `rapid-mlx`. ### Adoption history Rapid-MLX developed a fast release cadence and first-class integrations for coding agents and Python frameworks. By the 0.10 series it was distributed from PyPI and Homebrew core, while its documentation described dozens of model families, an alias catalog, and reproducible community hardware benchmarks. ### How it is used A typical workflow is `rapid-mlx serve ` followed by pointing an OpenAI-compatible client at `http://localhost:8000/v1`. Companion commands provide a terminal chat REPL, downloads and cache management, diagnostics, benchmarks, and setup templates for tools such as Codex, Claude Code, Aider, and Cursor. ### Why package nerds care Rapid-MLX packages the Apple-Silicon-specific MLX ecosystem into a familiar server-and-CLI interface, reducing model selection to stable aliases and preserving compatibility with the broad OpenAI client ecosystem. Its Homebrew, pip, uv, and self-contained installer paths make local LLM serving accessible without maintaining a bespoke Python environment. ### Timeline - 2026-02-25: Official GitHub repository created. - 2026-03-20: v0.3.0 published, the oldest retained GitHub release. - 2026: Command branding moved from vllm-mlx to rapid-mlx while compatibility entry points were retained. - 2026: Version 0.10.12 entered Homebrew core according to the official README. ### Related projects - Rapid-MLX builds on Apple MLX and MLX-LM concepts, consumes Hugging Face model repositories, offers vLLM-style flags, and acts as a local backend for OpenAI-compatible clients and coding agents. ### Sources - - - - ## Security Notes No matching local secret-handling manifest was found for rapid-mlx. Nucleus package metadata is still published here so future coverage has a stable package URL. ## Combined YAML source View the package source record on GitHub. [combined/rapid-mlx.yml](https://github.com/mxcl/pkgdb/blob/main/combined/rapid-mlx.yml) ## Sources - pkg.so package database - curated package history - pkgdb category and tag curation - cross-ecosystem install command graph