The right model for every prompt.

from openai import OpenAI

client = OpenAI(
    base_url="http://localhost:8000/v1",
    api_key="your-router-key",
)

resp = client.chat.completions.create(
    model="auto",  # let Jev decide
    messages=[{"role": "user", "content": "Refactor this to be async..."}],
)

print(resp.model)
# liquid/lfm2.5-24b-a2b
print(resp.model_extra["jev_router"]["routing_reason"])
# coding request, complex complexity; balanced strategy chose ...

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "http://localhost:8000/v1",
  apiKey: "your-router-key",
});

const resp = await client.chat.completions.create({
  model: "auto", // let Jev decide
  messages: [{ role: "user", content: "Refactor this to be async..." }],
});

console.log(resp.model);
// liquid/lfm2.5-24b-a2b

curl http://localhost:8000/v1/chat/completions \
  -H "Authorization: Bearer $ROUTER_KEY" \
  -H "Content-Type: application/json" \
  -H "x-jev-strategy: cheapest" \
  -d '{
    "model": "auto",
    "messages": [{"role": "user", "content": "Summarise this email"}]
  }'

# X-Jev-Router-Model: liquid/lfm3-1.7b
# X-Jev-Router-Strategy: cheapest