Inference — Intelligence, on request.

// OVERVIEW

THE INTERFACE

One endpoint. Many possibilities.

Chat, write code, understand images, or stream a response as it happens. All through an OpenAI-compatible API.

An etched celestial globe surrounded by orbital paths

FIG. 01 / A COMMON INTERFACE

A field of stars converging into a luminous point

FIG. 02 / MANY WAYS TO THINK

THE MODELS

A model for your next idea.

Explore the model catalog and choose what fits. Switch with a single parameter, or start with minimax-m3.

EXPLORE MODELS

ZERO PROMPT & RESPONSE RETENTIONYOUR FIRST REQUEST STARTS HERE ↓

// DEVELOPER GUIDE

From idea to first response.

BASE URL

https://inference.lulzx.space/api/v1

Pass any model from the catalog. Omit it to use minimax-m3.

1. Create an API token

No account is required. A token is displayed once and expires after 30 days.

curl -X POST https://inference.lulzx.space/api/token

Save the returned token value and send it with each request:

Authorization: Bearer YOUR_API_TOKEN

2. Chat completion

curl https://inference.lulzx.space/api/v1/chat/completions \
  -H "Authorization: Bearer $API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "messages": [
      {"role": "system", "content": "You are a helpful assistant."},
      {"role": "user", "content": "Explain entropy in two sentences."}
    ],
    "max_tokens": 256
  }'

Python

from openai import OpenAI

client = OpenAI(
    base_url="https://inference.lulzx.space/api/v1",
    api_key="YOUR_API_TOKEN",
)

response = client.chat.completions.create(
    model="minimax-m3",
    messages=[{"role": "user", "content": "Hello"}],
)

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

Streaming

Set stream to true. The response uses server-sent events.

{
  "messages": [{"role": "user", "content": "Write a short poem."}],
  "stream": true
}

Image input

Use an image URL in a chat message alongside a text instruction.

{
  "messages": [
    {
      "role": "user",
      "content": [
        {"type": "text", "text": "Describe this image."},
        {"type": "image_url", "image_url": {"url": "https://example.com/image.jpg"}}
      ]
    }
  ]
}

Models

Model	Context length
`dots-3-note-preview`	512,000
`flint-v1.0`	262,144
`gemma-4-26b-a4b-it`	262,144
`lfm-2.5-2.6b`	65,536
`nemotron-3-nano-omni-30b-a3b-reasoning`	256,000
`nemotron-3-ultra-550b-a55b`	1,000,000
`nemotron-3.5-content-safety`	128,000
`nemotron-3.5-lightning`	1,000,000
`nemotron-3.5-super-vl`	1,000,000
`north-mini-code`	256,000
`xing4.0-29b-a4b`	262,144

Endpoints

Method	Path	Description
POST	`/api/token`	Create an API token without an account.
GET	`/api/v1/models`	List the available models.
POST	`/api/v1/chat/completions`	Chat, vision, and streaming completions.
POST	`/api/v1/completions`	Legacy text completions.

Limits and errors

400: The request body is invalid.
401: The API token is missing, invalid, or expired.
429: Capacity or token-creation limit reached. Retry after the indicated interval.
5xx: The inference service is temporarily unavailable.

At most 10 inference requests are admitted per rolling minute. Token creation is limited to 3 per address per hour.

Usage statistics

Full dashboard: /statistics.

API tokens issued	5
Inference requests	29
Successful requests	20
Input tokens processed	1,433
Output tokens served	200
Total tokens processed	1,633
Counting since	2026-09-05T16:49:54.158044+00:00
Last request	2026-09-05T17:53:45.245858+00:00

Zero data retention

This gateway does not retain prompts or model responses. Only usage counts and operational metrics, such as model, status, latency, and token counts, are kept.

This policy covers the Inference gateway. Requests are sent to model providers for processing; their own data policies apply.

Experimental service. Availability and limits may change.