BASIS
Documentation

Documentation

The inference layer for the agent economy

Basis is a worker-powered, Base-native inference network. Try a model in the signed-in web console (sign-in required), or point any OpenAI client at the separate OpenAI-compatible API. Either way you pay in $BASIS credits (routing ETH, WETH, or USDC into $BASIS at payment time); contributor GPUs register as workers and can earn $BASIS for completed, verified jobs; and every settled job writes a verifiable inference receipt — a SHA-256-hashed accounting record, never the prompt text. $BASIS is the network token, live on Base; Bankr was the launch rail that brought it into being. These docs are honest about what is live versus pending — use /api/launch-status for current runtime state.

Developers

Start building

Point any OpenAI client at the Basis base URL — separate from the signed-in web console. Chat completions, streaming, models, auth and API keys, credits, and receipts.

Contributors

Run a worker

Register a contributor GPU with an EVM reward address, heartbeat, serve jobs the orchestrator routes to you, and earn $BASIS for completed work.

Token

Understand $BASIS

The token for inference credits, API usage, worker rewards, and settlement accounting — live on Base — with tokenomics, treasury, and Bankr as the launch rail.

Operators

Operate Basis

Configure the runtime, host the orchestrator, run the settlement keeper, wire durable persistence, and read what is live versus pending.

Quickstart

bash
# OpenAI-compatible. Point any OpenAI client at the Basis base URL:
#   https://basis.watch/api/v1
# Serves real completions when a backend is configured; otherwise a structured
# runtime_pending (503). Live runtime: GET /api/launch-status
curl -N https://basis.watch/api/v1/chat/completions \
  -H "Authorization: Bearer sk-basis-..." \
  -H "Content-Type: application/json" \
  -d '{"model":"basis-default","stream":true,
       "messages":[{"role":"user","content":"Summarize this for an agent."}]}'

API base URL

https://basis.watch/api/v1

Launch status

GET /api/launch-status

Network data

GET /api/data

Live vs pending — derived from configuration

Browse the docs

Agent-readable docs

The API is OpenAI-compatible and the docs are machine-consumable. A plain-text index of the network and these pages lives at /docs/llms.txt. Models are at GET /api/v1/models, receipts at GET /api/inference/receipts, and runtime state at GET /api/launch-status. Point your OpenAI client at https://basis.watch/api/v1.