mlx-serve alternatives
Tools that could take the place of mlx-serve, ranked by the stacks builders put them in rather than by stars alone.
Other runtimes in the runchr registry that do the work of mlx-serve — a tool somebody pairs with it ranks above a more popular stranger. Read the mlx-serve page or browse every runtimes.
Local AI server serving optimized LLMs on GPUs and NPUs via OpenAI-compatible
Run huge MoE models privately on your own GPUs with OpenAI-compatible API.
Local Qwen3.8-Flash-Next inference server for AMD Strix Halo with OpenAI
CLI proxy that compresses shell command output before it reaches an LLM.
Distributed AI compute engine for scaling ML training, tuning, and serving.
On-device inference engine for phones, wearables, and robots.
GPU kernel library for fast LLM inference: attention, GEMM, MoE, sampling.
OpenAI/Anthropic-compatible MLX inference server for Apple Silicon tuned
Local MLX inference server that serves open models to Claude Code via Anthropic
Fast local LLM inference engine for Apple Silicon with native MTP speculative
Run open models locally and serve an OpenAI-compatible API.
Run quantised models locally, with an OpenAI-compatible server.