API Documentation

Introduction

One OpenAI-compatible endpoint for 18 models — Claude, GPT, Grok, DeepSeek, Kimi, Mistral, Mercury, GLM, Qwen, Gemini and ERNIE — plus two Gemini TTS voices. Chat requests are auto-routed to a healthy provider; TTS returns audio bytes. Billing is per token for chat and per 1,000 characters for speech.

Quick Start

1. Create an API key in the Dashboard → API Keys.
2. Export it as an environment variable: NOVA_API_KEY
3. Make your first request with cURL, Python or Node below — any OpenAI SDK works.

curl https://YOUR-DOMAIN/v1/chat/completions \
  -H "Authorization: Bearer $NOVA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "glm-5.3-flash",
  "messages": [{ "role": "user", "content": "Hello, Nova!" }]
}'

from openai import OpenAI
import os

client = OpenAI(
    base_url="https://YOUR-DOMAIN/v1",
    api_key=os.environ["NOVA_API_KEY"],
)

response = client.chat.completions.create(
    model="glm-5.3-flash",
    messages=[{"role": "user", "content": "Hello, Nova!"}],
)

print(response.choices[0].message.content)

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://YOUR-DOMAIN/v1",
  apiKey: process.env.NOVA_API_KEY,
});

const response = await client.chat.completions.create({
  model: "gpt-6-luna",
  messages: [{ role: "user", content: "Hello, Nova!" }],
});

console.log(response.choices[0].message.content);

Authentication

All API requests are authenticated with a Bearer token in the Authorization header. Create and manage keys in the dashboard.

Authorization: Bearer $NOVA_API_KEY

Create chat completion

Request body

Pick a model

Every chat model uses the same endpoint — only the model ID changes.

Image input

Send an image URL (HTTPS or data:) as an image_url part. The model must be marked IMG in the model table.

curl https://YOUR-DOMAIN/v1/chat/completions \
  -H "Authorization: Bearer $NOVA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "gemini-4-argon",
  "messages": [{
    "role": "user",
    "content": [
      { "type": "text", "text": "Describe this image in one sentence." },
      { "type": "image_url", "image_url": { "url": "https://example.com/photo.jpg" } }
    ]
  }]
}'

Video input is only available on models marked VID. The accepted video part is provider-specific — contact support for the exact schema before sending video.

Response

{
  "id": "chatcmpl-9f2c84ab",
  "object": "chat.completion",
  "created": 1762830000,
  "model": "glm-5.3-flash",
  "choices": [
    {
      "index": 0,
      "message": { "role": "assistant", "content": "Hello! How can I help you today?" },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 12,
    "completion_tokens": 9,
    "total_tokens": 21
  }
}

Streaming

Set "stream": true to receive tokens as server-sent events as they are generated. The stream terminates with a data: [DONE] line.

data: {"id": "chatcmpl-9f2c84ab", "choices": [{"delta": {"content": "Hel"}}]}

data: {"id": "chatcmpl-9f2c84ab", "choices": [{"delta": {"content": "lo!"}}]}

data: [DONE]

stream = client.chat.completions.create(
    model="gpt-6-luna",
    messages=[{"role": "user", "content": "Write a haiku about GPUs."}],
    stream=True,
)

for chunk in stream:
    print(chunk.choices[0].delta.content or "", end="")

const stream = await client.chat.completions.create({
  model: "gpt-6-luna",
  messages: [{ role: "user", content: "Write a haiku about GPUs." }],
  stream: true,
});

for await (const chunk of stream) {
  process.stdout.write(chunk.choices[0]?.delta?.content || "");
}

Text-to-Speech

Two Gemini TTS models turn text into audio. The endpoint follows the OpenAI /audio/speech shape and returns raw audio bytes. Billing is per 1,000 input characters.

Request body

Pricing

1,000 characters costs $0.0055 (Lite) or $0.0080 (Full). Failed requests are not billed.

cURL

curl https://YOUR-DOMAIN/v1/audio/speech \
  -H "Authorization: Bearer $NOVA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "gemini-3.8-flash-tts",
  "input": "Hello from NovaCompute. This is a speech test.",
  "voice": "Kore"
}' \
  --output speech.wav

Use --output — the response body is audio, not JSON.

Python

import os, requests

response = requests.post(
    "https://YOUR-DOMAIN/v1/audio/speech",
    headers={"Authorization": f"Bearer {os.environ['NOVA_API_KEY']}"},
    json={
        "model": "gemini-3.8-flash-tts",
        "input": "Hello from NovaCompute.",
        "voice": "Kore",
    },
    timeout=60,
)
response.raise_for_status()

with open("speech.wav", "wb") as f:
    f.write(response.content)

Browser (demo only)

Never ship an API key in client-side code. For a quick local demo, proxy the request through your own server. The Playground already does this and plays the result.

const response = await fetch("/v1/audio/speech", {
  method: "POST",
  headers: {
    "Authorization": `Bearer ${apiKey}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    model: "gemini-3.8-flash-tts",
    input: "Hello from NovaCompute.",
    voice: "Kore",
  }),
});

const blob = await response.blob();
const audio = new Audio(URL.createObjectURL(blob));
audio.play();

Models

Pass any of these IDs in the model field. IMG and VID show accepted media input.

Prices are per 1M tokens unless marked per 1,000 characters. Free models cost $0.00.

Billing

Every account has a USD balance. Chat requests are charged per token at the model's input/output rate; TTS is charged per 1,000 characters. Failed requests from our side (5xx) are not billed. You can top up and see itemised usage in Dashboard → Billing.

Subscriptions

Sub plans replace per-token billing with a daily paid-token quota. Free models stay unlimited (fair use), and premium models include a smaller daily token allowance (shown on Models). PayPal is captured server-side and activates the plan automatically. Bank transfers auto-credit when a bank webhook is connected; otherwise support verifies the receipt before activation. Activation also adds bonus credit: Starter +$3, Pro +$5, Ultra +$12. After the daily quota is used, further paid requests bill from your balance instead of blocking.

Quotas reset daily. See Pricing for the full comparison and checkout.

What is billed

• Chat: input tokens + output tokens, at the rates in the model table.
• TTS: input characters only — $0.0055 or $0.0080 per 1,000 characters.
• Free models: $0.00, no balance required.
• Failed requests (5xx from NovaCompute): $0.00.

Errors

Errors return a consistent JSON shape:

{
  "error": {
    "type": "rate_limit_error",
    "code": "rate_limit_exceeded",
    "message": "Rate limit reached: 600 requests per minute."
  }
}

Rate limits

Every API key has a requests-per-minute cap and a daily request cap. Limits are checked before the upstream call; over-limit requests return 429 with Retry-After and rate-limit headers.

Need a guaranteed RPM or SLA? Contact sales for a committed-use tier.