Terms of Service
Last updated: 2026-05-11
1. The service
OpenLLM is a self-hosted gateway that proxies your requests to third-party LLM providers using the API credentials you connect on the Provider credentials page. The host does not resell inference capacity — every upstream call bills to the credential you provided.
2. Your responsibilities
- Maintain your own provider accounts and pay their bills. OpenLLM is not party to any contract between you and the upstream provider.
- Keep your API keys private. Anyone with a
sk-llm-…key can spend against your provider quotas through this gateway. - Comply with each upstream provider's terms — they govern the content of your prompts and outputs.
3. What we log
For every /v1/* call we record the model id, provider name, token counts, latency, status, and cost estimate to your private requests table for billing visibility. We do not log prompt contents or completions. See the Privacy Policy for the full list.
4. Termination
You can delete your account from the Settings page; the cascade wipes every row keyed to your user_id — API keys, credentials, request history, plugin embeddings, and plugin state. There is no soft-delete.
5. No warranty
The service is provided "as is" without warranty of any kind. Upstream provider outages, content moderation decisions, and billing disputes are between you and the provider — we cannot mediate them.