Quickstart

The gateway base URL is https://llms.party/v1. Authenticate with your virtual key:

export LLMS_PARTY_KEY="sk-..."

Keys are issued by your administrator. Each key is scoped to model groups and a budget; ask for a new key if you need a group it does not include.

1. Confirm the key

GET /v1/models is scoped to what the key may call. It is the cheapest round-trip and the first thing to try when a request fails.

curl -sS "https://llms.party/v1/models" \
  -H "Authorization: Bearer $LLMS_PARTY_KEY"

A 401 means the key is wrong or revoked. An empty data array means the key has no model groups.

2. First completion

curl -sS "https://llms.party/v1/chat/completions" \
  -H "Authorization: Bearer $LLMS_PARTY_KEY" \
  -H "content-type: application/json" \
  -d '{
    "model": "deepseek-flash",
    "messages": [{"role": "user", "content": "Explain an LLM gateway in one sentence."}]
  }'

import OpenAI from 'openai';
 
const client = new OpenAI({
  baseURL: 'https://llms.party/v1',
  apiKey: process.env.LLMS_PARTY_KEY,
});
 
const completion = await client.chat.completions.create({
  model: 'deepseek-flash',
  messages: [
    { role: 'user', content: 'Explain an LLM gateway in one sentence.' },
  ],
});
 
console.log(completion.choices[0]?.message.content);

import os
 
from openai import OpenAI
 
client = OpenAI(
    base_url="https://llms.party/v1",
    api_key=os.environ["LLMS_PARTY_KEY"],
)
 
completion = client.chat.completions.create(
    model="deepseek-flash",
    messages=[{"role": "user", "content": "Explain an LLM gateway in one sentence."}],
)
 
print(completion.choices[0].message.content)

Any OpenAI-compatible client works. Chat, streaming and vision use the OpenAI chat shape; video uses the video API.

3. Stream it

Set "stream": true and read SSE. Use streaming for requests that may think for more than a few seconds; an idle response can otherwise hit the gateway's proxy timeout before the first byte.

curl -N -sS "https://llms.party/v1/chat/completions" \
  -H "Authorization: Bearer $LLMS_PARTY_KEY" \
  -H "content-type: application/json" \
  -d '{
    "model": "deepseek-v4-pro",
    "stream": true,
    "messages": [{"role": "user", "content": "Plan a migration in three bullets."}]
  }'

4. Try it without a terminal

Chat playground

Same completion, streamed inline, against the live gateway.

Video generation

Create a MiniMax H3 or BytePlus Seedance job and watch it poll.

Errors & limits

What each status code means and where budgets bite.