# MLX Docs

The reference for building on MLX and for checking it. Run models on your own Mac, find the best AI for it in the AI App Store, read the network, sign transactions, and verify the ledger. Every number here is read from the code that enforces it.

## Start here

:::cards
- [Quickstart](/start/quickstart) Three paths: run a model on your Mac, read the network, or earn with a Mac.
- [Concepts](/start/concepts) Wallets, MLX and USDC, dedicated and free seats, Macs, jobs and blocks in one page.
- [Local AI](/local/overview) mlx-lm's server runs an open model on your own Mac behind an OpenAI-style chat API.
- [AI App Store](/store/overview) A curated catalogue of AI apps, tools and open models for the Mac and iPhone, with a signed API and the MLX Score.
- [Network API](/network/overview) Public JSON for blocks, transactions, wallets, seats, the curve and compute rewards. No key needed.
:::

## Build

:::cards
- [Transactions](/transactions/overview) Keys from the 24-word phrase, canonical JSON, Ed25519 signatures, every transaction type.
- [MLX Node](/node/overview) How a Mac talks to the network: the signed worker protocol, remote commands and links.
- [Compute API](/compute/overview) Paid AI jobs on seat-holding Macs, paid in USDC. Pilot.
- [Rate limits](/resources/rate-limits) The real limits on every public endpoint, read from the code.
:::

## Check

:::cards
- [Integrity](/integrity/overview) The transparency log, signed checkpoints, inclusion and consistency proofs, per-wallet state proofs, anchors on Solana and Bitcoin.
- [Protocol](/protocol/overview) The exact rules: seats, the hourly mining formula, lock-ins, the curve, how jobs are checked.
- [Parameters](/protocol/parameters) Every protocol constant in one table, with where it lives in the code.
- [The work](/network/compute) What share of the Macs' work is paid, and how often Macs with different chips agree.
:::

> [!NOTE]
> These docs are in English. For guides in 31 languages (earning on a Mac, the best Mac for local LLMs), see [mlx.app/learn](https://mlx.app/learn). For the reasoning behind the design, see the [whitepaper](https://mlx.app/whitepaper).