whereismyai
oMLX
oMLX is an inference server for macOS that provides an OpenAI and Anthropic compatible API for Apple Silicon using MLX.
oMLX is a native macOS inference server built on MLX designed for Apple Silicon. It optimizes performance for coding agents by persisting KV cache blocks to SSD, allowing the system to restore prefixes in milliseconds rather than recomputing them. This approach prevents the long wait times typically associated with agent interactions that frequently invalidate caches. The server provides a drop-in API compatible with tools like Claude Code, OpenClaw, and Cursor. It supports existing local setups by reading from the Hugging Face cache and LM Studio folders.
oMLX at a glance
- Best for: Software engineer
- Runs on: macOS
- Pricing: Open source (Apache 2.0 license)
- Released: February 2026
- Category: AI coding tool
- Website: omlx.ai
Key features
- Paged SSD KV caching
- Continuous batching
- OpenAI and Anthropic compatible API
- Integration with Claude Code, OpenClaw, and Cursor
- Support for Hugging Face cache and LM Studio folders
- Native Apple Silicon optimization
Use it for
- Reducing time-to-first-token for local coding agents
- Running local LLM inference on macOS
- Connecting local models to developer tools like Cursor
Companions
Tools that articles about oMLX recommend using with it, each with its job.
- Context7 — Provides up-to-date documentation for AI agents and code editors.
- Ollama — Run local models at zero marginal cost, with code kept on your machine.
- OpenRouter — Access over 500 models and 80 providers through one unified API.
- Daytona — Runs sessions in isolated sandboxes synced to git branches.
- Composio — Equips AI agents with tools, managed user auth, and sandboxes across 1,500+ apps
- Oh My OpenCode Slim — Lightweight multi-agent orchestration and background tasks.
Alternatives to oMLX
- MTPLX — It runs local models twice as fast via native MTP speculative decoding.
- Rapid-MLX — Run 250 MLX-native models locally on Apple Silicon with high inference speeds.
- OpenCode — Works with 75+ model providers, local models, and existing subscriptions
- OmniRoute — Routes requests across 352 providers with automatic fallback and cost-based selection.
- OpenMonoAgent — Runs locally on your own hardware with no recurring subscription fees.
- MLX LM — Runs LLMs and fine-tuning natively on Apple silicon hardware.
- ODS — Automates complex local AI infrastructure setups with one command.
- Mesh LLM — Run large local models across multiple machines through one OpenAI-compatible endpoint.
- AirLLM — Runs massive 70B+ parameter models on small 4GB consumer GPUs.
- llm-d — Deploys production-grade LLM serving on existing hardware clusters.
- Pulsar — It is the fastest inference engine for Qwen models on Apple Silicon.
- LMDeploy — Accelerates inference with high-performance engines and quantization tools.
All oMLX alternatives