RouterBusDrop-in OpenAI endpoint

Point an existing OpenAI client at one base URL. Metered per token, and every call is attributable in the log.

🔌

One endpoint

Standard /v1/chat/completions, with streaming. Change a base URL and a key; leave the rest of your code alone.

🔄

openai-python, openai-node, or a plain curl — anything that speaks the OpenAI protocol works unchanged.

🧩

Claude, GPT / Codex and Gemini behind a single address — switching models is one string.

🔍

Every request lands in the usage log with its own token counts and cost, against the API key that made it.

💰

No subscription, no seat fees, no monthly minimum. Live per-model prices are published in Model Square.

Describe a situation, ask a few questions, and get judgements back with probabilities — for routing, classifying and prioritising requests.

Learn about Jev