AI Gateway

One API key. Every AI modality. Unified under a single endpoint.

Overview

AI Gateway is LiteStartup's unified multimodal AI API. Instead of managing separate provider accounts for LLM, image, video, and audio — use one API key for all of them with unified billing, rate limiting, and intelligent model routing.

Capabilities

Text & Language

Audio

Image

Video

Quick Start

1. Get your API key

Go to Settings > API Keys in your dashboard and create a new key. This key works for all AI Gateway endpoints.

2. Call the API

All modalities use the same unified endpoint — just change the model field:

curl https://api.litestartup.com/client/v2/ai/generations \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-v3",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

3. Get results

• Sync — LLM chat, TTS, OCR, and transcription return results immediately.
• Async — Video generation and digital human return a task_id. Poll /client/v2/ai/tasks/{id} for status.

API Endpoints

OpenAI-Compatible Mode (LLM)

For LLM chat, you can use the standard OpenAI SDK with our base URL:

from openai import OpenAI

client = OpenAI(
    api_key="YOUR_API_KEY",
    base_url="https://api.litestartup.com/client/v2/ai"
)

response = client.chat.completions.create(
    model="deepseek-v3",
    messages=[{"role": "user", "content": "Hello!"}]
)
print(response.choices[0].message.content)

This endpoint is fully compatible with OpenAI's Chat Completions API — same request/response format, same SDKs.

Note: /client/v2/ai/chat/completions only supports LLM models. For image, video, audio and other modalities, use the unified /client/v2/ai/generations endpoint.

The model field determines which capability is invoked. See the API Reference for full request/response documentation.
For LLM-only usage with existing OpenAI SDKs, see the OpenAI-Compatible Mode section above.

Tip: Try it interactively in the Playground within your dashboard.

Model Selection

Specify the model you want directly in the model field:

{"model": "deepseek-v3", "messages": [...]}
{"model": "seedream-v5.0-lite-t2i", "input": {...}}
{"model": "wan-2.7-t2v", "input": {...}}

Available models are listed in the Playground. Intelligent routing features are planned for a future release.

Sync vs Async

Synchronous endpoints return results in the HTTP response:

• LLM Chat, OCR, Speech to Text, Text to Speech

Asynchronous endpoints return a task_id immediately:

• Video Generation, Image to Video, Digital Human

For async tasks, poll GET /client/v2/ai/tasks/{task_id} until status is completed, then retrieve the output URL from the response.

Billing

AI Gateway usage consumes AI credits from your plan's monthly quota. Different capabilities consume different amounts based on processing complexity. See the Pricing page for details.