Package Exports

@c0mpute/worker
@c0mpute/worker/dist/index.js

This package does not declare an exports field, so the exports above have been automatically detected and optimized by JSPM instead. If any package subpath is missing, it is recommended to post an issue to the original package (@c0mpute/worker) to support the "exports" field. If that is not possible, create a JSPM override to customize the exports field for this package.

Readme

c0mpute-worker

Native CLI worker for the c0mpute.ai distributed inference network. Runs LLM inference via ollama and connects to the orchestrator via Socket.io.

Quick Start

Install ollama and make sure it's running (ollama serve)
Run the worker:

npx @c0mpute/worker --token <your-token>

On first run, the worker will automatically:

Pull the base model (~17GB download)
Create a custom c0mpute-max model with optimized settings
Run a speed benchmark
Connect to the orchestrator and start serving jobs

How It Works

Verifies ollama is running locally
Pulls and configures the model (Qwen 3.5 27B abliterated)
Runs a speed benchmark to measure your hardware
Connects to the c0mpute.ai orchestrator via WebSocket
Accepts and processes inference jobs, streaming tokens back in real time

Capabilities

Thinking — model uses chain-of-thought reasoning with <think> tags
Vision — accepts images (base64) alongside text messages
Tool calling — model can invoke tools (web search, etc.) defined by the orchestrator
Uncensored — abliterated model with no content restrictions
Long context — 256K context window

Options

--token <token>   Authentication token from c0mpute.ai (required)
--url <url>       Orchestrator URL (default: https://c0mpute.ai)
--benchmark       Run benchmark only, then exit
--version         Show version
--help            Show help

Requirements

Node.js 18+
ollama installed and running
GPU with 20GB+ VRAM recommended (NVIDIA RTX 3090/4090, Apple Silicon 32GB+)
~17GB disk space for the model

Default Model

Qwen 3.5 27B Abliterated — an uncensored 27B parameter model with 256K context window, vision support, and thinking capabilities.

Architecture

The worker delegates all inference to ollama's local HTTP API. This means:

No CUDA/Metal build issues — ollama handles GPU acceleration
Easy model management — ollama pulls and caches models
Automatic GPU detection — ollama picks the best backend for your hardware

The worker is a dumb relay — it passes tool definitions to the model and relays tool calls back to the orchestrator for execution. Tools are defined and managed server-side.

Earnings

Workers earn credits for completing inference jobs. Earnings are based on tokens generated and your hardware tier. Check your earnings at c0mpute.ai.