Lumen API vs the field
Most "routers" are marketplaces or proxies — they unify billing and let you choose the model. Lumen makes the choice for you, per request, with a quality floor. Here's the honest landscape.
| Capability | Lumen API | OpenRouter | Martian | Not Diamond | LiteLLM | Requesty |
|---|---|---|---|---|---|---|
| Single OpenAI-compatible endpoint | Yes | Yes | Yes | Routing only | Yes | Yes |
| Auto-picks model per request | Yes — difficulty engine | You pick | Yes | Yes | You configure | Rules |
| Quality floor per task class | Yes | No | Cost/quality knob | Yes | No | No |
| Mid-flight provider failover | Yes | Yes | Limited | N/A | Yes | Yes |
| Geopolitical provider filter (no CN/RU) | Built-in default | User must curate | User curate | N/A | User curate | User curate |
| Hash-chained per-request audit log | Yes | Activity logs | No | No | Self-host logs | Logs |
| Dollars-saved reporting | Per response | No | Estimate | No | No | Spend only |
| BYO provider keys | Pro+ | Yes | Some | Yes | Yes (self-host) | Yes |
| Battle-tested at scale before launch | 40+ apps, tens of M calls | Yes | Yes | Yes | OSS, large | Growing |
The honest take: OpenRouter and Requesty are great marketplaces — Lumen is happy to route through model access you already have. LiteLLM is a great self-hosted proxy if you want to wire your own routing logic. Martian and Not Diamond are the closest peers — both also do automatic routing. Lumen's differentiators are (1) the geopolitically-filtered provider set as a default, (2) the dollars-saved-per-response transparency, and (3) the hash-chained audit log out of the box. We're not the only autonomous router — we're the one that ships compliance and transparency by default.