The inference API developers actually enjoy

import Nuvora from "@nuvora/sdk";

const nv = new Nuvora();

const stream = await nv.chat.create({
  model: "auto", // cheapest model that passes your evals
  messages: [{ role: "user", content: "Summarize this PR" }],
  stream: true,
});

for await (const chunk of stream) {
  process.stdout.write(chunk.delta);
}

from nuvora import Nuvora

nv = Nuvora()

stream = nv.chat.create(
    model="auto",  # cheapest model that passes your evals
    messages=[{"role": "user", "content": "Summarize this PR"}],
    stream=True,
)

for chunk in stream:
    print(chunk.delta, end="")

curl https://api.nuvora.ai/v1/chat \
  -H "Authorization: Bearer $NUVORA_KEY" \
  -d '{
    "model": "auto",
    "stream": true,
    "messages": [
      { "role": "user", "content": "Summarize this PR" }
    ]
  }'