# unbound.ai

> An OpenAI-compatible AI API for cybersecurity research, code review and agent workflows.

Canonical production origin: https://unbound-me.com. This response is served by the production environment. Check /public/runtime for registration and demo-payment availability. Check deployment status before running examples; documentation is not an uptime guarantee.

## Public documentation

- [Public Test API](https://unbound-me.com/docs/test-api.md)
- [Coding app setup](https://unbound-me.com/docs/opencode.md)
- [GLM, DeepSeek & MiniMax API Models](https://unbound-me.com/models.md)
- [AI API Pricing per Million Tokens](https://unbound-me.com/pricing.md)
- [API & MCP Documentation](https://unbound-me.com/docs.md)
- [AI API Quickstart & Authentication](https://unbound-me.com/docs/api.md)
- [Unbound MCP Setup & Agent Login](https://unbound-me.com/docs/mcp.md)
- [Token Caps, Spend Limits & Billing](https://unbound-me.com/docs/billing.md)
- [Terms of service](https://unbound-me.com/terms.md)
- [Privacy policy](https://unbound-me.com/privacy.md)
- [Acceptable use](https://unbound-me.com/acceptable-use.md)
- [Payments, refunds & complaints](https://unbound-me.com/refunds.md)
- [Cookies & browser storage](https://unbound-me.com/cookies.md)
- [Security reporting](https://unbound-me.com/security.md)
- [Contact unbound.ai](https://unbound-me.com/contact.md)

## Available model names and token rates

| Model | API ID | Input USD / 1M | Output USD / 1M | Cached input USD / 1M |
| --- | --- | ---: | ---: | ---: |
| GLM-5.3-753B-Unbound | `glm-5.3-753b-unbound` | 1.2 | 2.4 | 0.18 |
| GLM-5.3-Flash-320B-Unbound | `glm-5.3-flash-320b-unbound` | 0.144 | 0.6 | 0.036 |
| DeepSeek-V4.1-Flash-552B-Unbound | `deepseek-v4.1-flash-552b-unbound` | 0.144 | 0.54 | 0.03 |
| DeepSeek-V4.1-Flash-552B | `deepseek-v4.1-flash-552b` | 0.012 | 0.096 | 0.0024 |
| MiniMax-M3.1-Flash-428B | `minimax-m3.1-flash-428b` | 0.024 | 0.096 | 0.0024 |

Prices are USD per one million tokens, not per request or per model download. Cached input rates apply only when the provider reports cache usage. No benchmark advantage or unlimited/free API access is promised here.