> For the complete documentation index, see [llms.txt](https://docs.brainnetwork.app/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.brainnetwork.app/readme.md). # The crowd is the GPU BRAIN turns ordinary computers into one AI compute network. This book explains exactly how, what is real today, and what happens as the crowd grows. Every laptop, desktop and gaming PC on earth has a GPU that sits idle most of the day. Meanwhile AI inference runs in a handful of datacenters that are expensive to build, booked out, and owned by very few companies. BRAIN connects those two facts. Open a browser tab and your GPU is measured, verified and put to work. Run a small agent on a machine with an NVIDIA card and it serves open-weight language models to developers. Send a request to an OpenAI-compatible endpoint and it is routed to whichever machine can answer it cheapest and prove that it did. This book is the full account of how that works: the routing formula, the verification rules, how the hourly SOL pool is split, where every coin comes from and goes to, and what the network becomes as it grows from a few hundred devices to a provider nobody owns. {% hint style="info" %} **How to read the labels.** Every component of BRAIN carries one of three tags, and this book uses them the same way the code does. * **LIVE** — deployed in production, exercised by tests, producing figures from real records. * **DEMO** — works end to end on a labelled mock. Never mixed with LIVE figures. * **PLANNED** — an interface or stub exists so the shape is visible. It does nothing yet. If a number is not measured, it is shown as **UNKNOWN**, not estimated. {% endhint %} ## The network right now Numbers move; the page that reads them live is [brainnetwork.app/network](https://brainnetwork.app/network) and the machine-readable source is [`/api/stats`](https://brainnetwork.app/api/stats). The snapshot below was taken on **7 October 2026 at 19:25 UTC**, three days after launch, and is included so you can see what the words in this book refer to. | | | | ------------------------------------- | ------------------------------------------------------------------------------------------------------------------------- | | Devices online | 619 | | Devices that have joined since launch | 7,092 | | Dedicated GPU nodes registered | 4 (RTX 4060, RTX 3070, RTX 3060, RTX 3050 laptop) | | Work units verified | 2,259 | | Verified compute units | 2,660,446 | | SOL paid to contributors since launch | 11.10 | | Payout wallet | [`7TSAzUKLyCPb5AqzXUPMy4mTwVj4sdng2Gu2omSsaxVF`](https://solscan.io/account/7TSAzUKLyCPb5AqzXUPMy4mTwVj4sdng2Gu2omSsaxVF) | ## Where to start | If you are | Start here | | ------------------------------------ | ----------------------------------------------------------------------------------------------------------- | | Curious what this is | [What BRAIN is](/the-idea/what-brain-is.md) | | Going to open a tab and earn | [Run a browser node](/participate/run-a-browser-node.md) | | Running an NVIDIA machine | [Run a GPU node](/participate/run-a-gpu-node.md) | | Building with the API | [Use the API](/participate/use-the-api.md) | | Checking whether any of this is real | [Real vs simulated](/trust/real-vs-simulated.md), then [Track everything](/participate/track-everything.md) | | Thinking about where this goes | [The shape of the network](/scale/the-shape-of-the-network.md) | ## One paragraph for the impatient You ask. BRAIN finds the cheapest computer that can answer, sends you the answer with a receipt, and pays the computer that did the work. Your computer can be one of those computers. Only verified work earns. Everything is on-chain or in a public ledger you can read. {% hint style="warning" %} BRAIN is experimental software, three days old at the time of writing. It has already had two production outages, both described in the [incident log](/trust/incident-log.md). Rewards are not guaranteed and nothing here is investment advice. {% endhint %} --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters: ``` GET https://docs.brainnetwork.app/readme.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `build a script that syncs our docs to a CMS` lets GitBook tailor the answer to that use case. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.