---
title: "Boundless Platform"
description: "Boundless Platform: authentication, endpoints, examples, limits and agent integrations"
canonical: "https://inference.boundless.network/docs"
last-updated: "2018-10-20T01:46:40.000Z"
---

# Boundless Platform

An inference platform for open-weight models, with prepaid credits and exact per-request costs.


Boundless serves open-weight models over an OpenAI-compatible API, sold on prepaid credits. Every response carries its exact token counts, so any charge on your account can be recomputed from the rate card. We start with the workload, then tune the model and the serving stack around it.

## How it works

> If you are an agent, [try the platform without an account](/docs/quickstart#try-it-without-an-account).

1. **Sign up and get approved.** We activate accounts by hand.
2. **Create an API key** in [API Keys](/api-keys).
3. **Point your OpenAI SDK at our base URL.** Existing code works without changes.
4. **Spend against your credit balance.** Every request returns its token usage and debits the balance exactly. When credits reach zero, requests stop.

## Zero data retention

Your prompts and outputs are never stored or trained on. We keep what billing needs: token counts, model, timestamp, and cost. That is what the usage page shows you, and it is enough to recompute any charge yourself. Request and response bodies are never written to a log or a database.

[Section 3 of the terms](/terms) says this in full, and sets out the limits on what we can promise about the third-party infrastructure we run on.

## Where to go next

- [Quickstart](/docs/quickstart): from API key to first streamed token.
- [Coding Agents & Harnesses](/docs/coding-agents-and-harnesses): point Codex CLI, Claude Code, omp, Hermes and other harnesses here.
- [API Compatibility](/docs/api-compatibility): the endpoints we serve.
- [Credits & Billing](/docs/billing): buying credits, automatic top-ups, per-key budgets, and the hard stop at zero.
- [Pricing](/docs/pricing): the models we serve and the rate card.
- [Rate Limits](/docs/rate-limits): the throughput cap on your team, and how to raise it.
- [Support](/support): how to reach us, and what to send with a question.

## Machine-readable developer resources

- [OpenAPI 3.1 specification](/openapi.json): typed endpoints, authentication, permissions, errors, and response headers.
- [Authentication guide](/auth.md): the exact human-issued API-key flow and its automation boundaries.
- [MCP manifest](/.well-known/mcp.json): the product MCP server and Streamable HTTP endpoint.
- [llms.txt](/llms.txt): service summary, fit guidance, documentation, and every agent-facing resource.