selan.aiDocs
The models your team asks for
Claude, GPT, Kimi, DeepSeek, Qwen and GLM run through one endpoint, on your company's own accounts, with one usage record across all of them.
What you get
Six model families
Name any other model and OpenRouter reaches it: Llama, Mistral, Grok, Gemini, Nova, Command and several hundred more.
Where they come from
Three subscriptions
These three are capacity you have already bought. Between them they cover every model above, and each one has something behind it for when it runs out.
Seats your team already has. Connect as many as you hold and Selan spreads the load across them.
- Claude API uses your Anthropic key and bills per token. A key that starts refusing is benched rather than retried into a lockout.
- Google Vertex runs the same models on your own Google invoice.
- nexos.ai is a third route to Claude, on a separate contract.
The plan you already pay for, metered into the same usage record as everything else.
- OpenRouter serves GPT per token when the plan is capped.
One flat monthly price for the open-weight models. No per-token bill at all.
- OpenRouter has the same open-weight models per token, plus the long tail nobody wanted a separate contract for.
Selan tries subscriptions first. It spends the capacity you already own before it spends anything metered, and falls through to a key only when the seats are capped. One usage record covers both.
Ollama and ChatGPT answer only when you name them. Selan never substitutes anything silently.