What is talos?

Talos is a client program that lets someone share their computer's GPU with the Talos network in exchange for earnings. Once installed and paired with a Talos account using a one time code, it serves AI inference jobs from the network through a locally running Ollama installation, reports how long it stays connected, and earns a share of the revenue generated by real usage on the network. The Talos web dashboard and this client only communicate over the network through a pairing code, followed by jobs and status updates sent over a WebSocket connection. To run it, a person needs Python 3.9 or newer and Ollama running locally with at least one model already downloaded. An NVIDIA GPU is recommended and detected automatically, though the client also works on a CPU. After installing with pip, the user pairs the machine using a code from their Talos dashboard, either by typing it in when prompted or passing it directly as a command line option along with the server address. Running the worker opens a small local dashboard in the browser showing live status and a slider that controls how much of the machine's capacity to offer to the network, expressed as a value between zero and one that affects concurrency rather than a literal power percentage. While connected, uptime accumulates and earnings are credited per job served, both visible on the user's Talos dashboard. A status command shows the current configuration, detected GPU, and available models and jobs. The same repository also serves a second, unrelated audience: developers who simply want to use a feature called talos-auto inside their code editor without sharing any GPU at all. For that use case there is no need to run the worker software, since a separate setup tool or manual configuration guides exist for editors like Cursor, VS Code, Claude Code, JetBrains IDEs, Zed, and Aider, each pointing the editor at a hosted gateway instead of a local machine.