SMALL PAPER
DOCS TELEGRAM CHAT LOG IN

Unrestricted OpenAI-Compatible API

Build what other LLMs refuse.

Abliterated open-weight and frontier models behind one OpenAI-compatible API — built for reverse engineering, security research, red-teaming, and workloads standard guardrails block.

Base URL

https://api.smallpaper.io/v1

  • Speedfrontier models, agent-fast
  • Cheapup to 91% off list price
  • Uncensoredsame models, no refusals
Model Prices ↓
smallpaper-cliInstall

One API. Full control.

  • Abliterated open-weight and discounted frontier models.
  • Drop-in OpenAI /v1 for smallpaper-cli, opencode, or SDKs.
  • Streaming tool calls, reasoning budgets, and 1M context.
  • 20-digit key in one click — crypto, no sign-up.

In the chat

Make anything you can imagine

Draw, edit and combine pictures, or generate short clips in the web chat, billed per unit from your balance or pass.

A black and white cat with a black mask and white muzzle, sitting on a sunlit window sill image

Oreo, a black and white cat, sitting on a sunny window sill

Ideogram 4 · text → image

The same cat lying in a grey fabric hammock on a cat tree, seen from above image reference

put Oreo in this cat-tree hammock

Seedream 4.5 · edit & combine

video · beta

Oreo gets up and jumps down to the floor

MiniMax H3 MAX · 4–15s · 480p/768p

Per-unit prices show live in the model tables below.

Uncensored

Uncensored versions of these models, with their refusals removed. Not the stock releases.

ModelInputCached inputOutput You saveContext
Kimi K3.1 Pro HeavyIncluded on the Origami plan; otherwise paid from your balance.Kimi K3.1 Abliterated MXFP4$8.50input$8.50$0.85$20.50—1M
GLM 5.3 HeavyIncluded on the Origami plan; otherwise paid from your balance. ZDRZero Data Retention: prompts and completions are never stored by the provider.GLM 5.3 Abliterated FP8$4.725input$4.725$0.75$7.50—1M
GLM 5.3 [Cyber]Experimental: may refuse but cheaper$1.90input$1.90$0.35$5.60—1M
DeepSeek V4.1 Flash 1017The abliterated variant$0.50input$0.50$0.035$1.90—1M
GLM 5.3 Flash DefaultGLM 5.3 Flash Abliterated$0.45input$0.45$0.20$0.95—1M
Zeus500 tokens/s beast (experimental: may refuse)$0.85input$0.85$0.032$2.50—1M

Cheap

Billed strictly per $/M tokens, cached pricing passed through at zero loss.

ModelInputCached inputOutput You saveContext
GPT 6.1 Sol Best dealOpenAI's fast frontier release$0.32$2.00$0.018$0.10$1.12$10.00−88%1M
Gemini 4 CarbonGoogle's heavyweight. Slower to start, built for the hard problems$0.45$2.00$0.03$0.10$2.10$10.00−79%1M
Gemini 4 ArgonGoogle's heavyweight reasoning model$0.48$2.00$0.03$0.10$2.30$10.00−77%1M
DeepSeek V4.1 Flash 1017DeepSeek's fast one. Cheap enough to leave running$0.06$0.15$0.0015$0.003$0.24$0.60−60%1M
Kimi K3.1 ProMoonshot frontier model$0.35$2.00$0.04$0.20$1.45$10.00−85%1M
GLM 5.3 DefaultBalanced uncensored GLM 5.3 release$0.15$1.40$0.036$0.26$0.45$4.40−90%1M
Zeus [CHEAP]Fast uncensored reasoning model$0.10$0.01$0.45—1M
Claude Opus 5.5 Off saleTemporarily paused; back on sale shortly.Anthropic frontier model$0.95$4.00$0.054$0.20$4.75$20.00−76%1M

Live · repriced

Ideogram 4 · Seedream 4.5 Ideogram · ByteDancedraws, edits and combines in all model chats $0.025 / image

02 Pay and sign in

Don't believe us? All you need is $2

xRocket

Top up with crypto.

$2 to start

Prepaid, credited 1:1. Balance never expires.

Your number is your key

20 digits, no email or password. Inspired by Mullvad VPN.

A key per machine

Pair each device, revoke any one. Your number never leaves the browser.

No logs

We keep time, model, size and cost — never your prompts or replies.

or pay once

Passes: one fixed price.

A pass pays first for the models it covers; your wallet pays the rest.

Starter
◆◇◇◇$15

Pass · 30 days001/004

For trying it properly.

  • Chat messages≈ 220 a day
  • Agentic coding≈ 5.9 h a day
  • Heavy output≈ 3.2 h a day
  • Long-context work≈ 2.9 h a day

Estimates on GPT 6.1 Sol at today's rates.

HEAVY models: Origami only

Get a pass
Vibecoder
◆◆◇◇$40

Pass · 30 days002/004

A working day, every day.

  • Chat messages≈ 580 a day
  • Agentic coding≈ 16 h a day
  • Heavy output≈ 8.5 h a day
  • Long-context work≈ 7.7 h a day

Estimates on GPT 6.1 Sol at today's rates.

HEAVY models: Origami only

Get a pass
Hyperscaler
◆◆◆◇$99

Pass · 30 days003/004

Agents, running.

  • Chat messages≈ 1,450 a day
  • Agentic coding≈ 39 h a day across 6 agents
  • Heavy output≈ 21 h a day across 6 agents
  • Long-context work≈ 19 h a day across 6 agents

Estimates on GPT 6.1 Sol at today's rates.

HEAVY models: Origami only

Get a pass

For heavy agents and teams

Origami
◆◆◆◆$500

Pass · 30 days004/004

Nothing held back.

  • Chat messages≈ 7,100 a day
  • Agentic coding≈ 191 h a day across 10 agents
  • Heavy output≈ 104 h a day across 10 agents
  • Long-context work≈ 93 h a day across 10 agents

Estimates on GPT 6.1 Sol at today's rates.

+ HEAVY models included

Get a pass

Getting started

Three steps.

Any OpenAI-compatible client works.

01Create an account

A 20-digit number, generated in the browser. No email, no password. It is also your API key.

02Fund it

Pay by crypto. Credited 1:1 in USD, and balances never expire.

03Point your client at it

One command, then any model we serve.

  1. 01
    Paste and run it in a terminal.
    # macOS / Linux
    curl -fsSL https://smallpaper.io/install.sh | bash
    # Windows · PowerShell
    irm https://smallpaper.io/install.ps1 | iex
  2. 02
    It writes the opencode provider config and pairs this machine — a separate, revocable key.
  3. 03
    Start opencode and pick any model we serve.

Full opencode guide

A terminal coding agent for every model here. One file: it installs itself, pairs this machine and keeps itself up to date.

Windows warns once because the file is unsigned: More info, then Run anyway. No build yet for Intel Macs, arm64 Linux or Windows on ARM.

macOS and Linux, from a terminal:

# macOS (Apple silicon)
curl -fsSL -o smallpaper \
  https://smallpaper.io/cli/download/smallpaper-aarch64-apple-darwin \
  && chmod +x smallpaper && ./smallpaper install
# Linux (x86_64, any distro)
curl -fsSL -o smallpaper \
  https://smallpaper.io/cli/download/smallpaper-x86_64-unknown-linux-musl \
  && chmod +x smallpaper && ./smallpaper install

Install guide

from openai import OpenAI

client = OpenAI(
    base_url="https://api.smallpaper.io/v1",
    api_key="41920837556120947381",   # your account number
)
r = client.chat.completions.create(
    model="glm-5.3-flash-abliterated",
    messages=[{"role": "user", "content": "..."}],
)
import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.smallpaper.io/v1",
  apiKey: "41920837556120947381",
});
const r = await client.chat.completions.create({
  model: "glm-5.3-flash-abliterated",
  messages: [{ role: "user", content: "..." }],
});
curl https://api.smallpaper.io/v1/chat/completions \
  -H "Authorization: Bearer 41920837556120947381" \
  -H 'content-type: application/json' \
  -d '{"model": "glm-5.3-flash-abliterated",
       "messages": [{"role": "user", "content": "..."}]}'