About
Large language models. One gateway.
Prefill Systems is an inference provider starting with Qwen3.8-27B: an OpenAI-compatible API, transparent pricing, and a zero data retention policy.
Why we exist
We make language-model inference straightforward to integrate and operate. Developers can request direct API access now; distribution through OpenRouter is pending approval.
Direct access
Approved beta customers can generate API keys, track token usage, and review estimated charges in their dashboard. Request access with your expected workload.
Our service
| Fact | Status |
|---|---|
| Public gateway | Live, monitored, controlled beta |
| OpenRouter integration | Pending approval |
| Inference data policy | Zero data retention — no persistent prompt or completion storage |
| SLA / uptime commitment | None yet — see the status page |
Built around your workflow
Use familiar OpenAI-compatible clients, stream responses as they arrive, and manage access through account-scoped API keys. See Qwen3.8 API pricing, use the Python guide, or request direct access.
Principles
- Transparent pricing. Per-token rates and account-level usage reporting make costs easy to follow.
- Zero data retention. No persistent storage of prompts or completions. Necessary operational metadata is retained separately.
- Controlled access. Create and revoke API keys from your dashboard, with usage attributed to your account.