OpenAI-compatible API · pay in crypto

Models without the refusals.
A key in a minute.

Get an API key in Telegram, top up with crypto and paste the address into Cline, Roo, Cursor, Aider or your own code. You pay only for the tokens you use.

Get a key in Telegram Read the docs
$ curl https://api.palantir-gpt.com/v1/chat/completions \
    -H "Authorization: Bearer sk-..." \
    -d '{"model": "glm-5.3-flash", "messages": [...]}'

{ "model": "glm-5.3-flash",
  "choices": [{ "message": { "role": "assistant", "content": "..." } }],
  "usage": { "prompt_tokens": 12, "completion_tokens": 96 } }
from $1minimum top-up
2 minutesfrom /start to the first request
5models in the catalog
USDT & morepaid via Crypto Bot

Why it is convenient

Compatible

The Chat Completions protocol, like OpenAI. Swap the address and the key; everything else in your client stays the same.

All in Telegram

Keys, balance, top-ups and support in one bot. No website sign-up.

Honest billing

You are charged for the tokens actually used. The price per 1M tokens is listed below, with no subscription.

Several keys

Up to five keys per account: one per project or client, and any of them can be revoked.

Getting started

  1. 1

    Open the bot

    Press /start in Telegram and accept the terms.

  2. 2

    Top up and create a key

    Top up in crypto, then create a key in the API section.

  3. 3

    Paste into your client

    Base URL and key; pick the model by the name from the price table.

Models and pricing

USD per 1 million tokens. Cache means a re-sent context prefix.

glm-5.3-flashdefault

Fast and cheap, the default

input
$0.20
output
$0.60
cache
$0.20
glm-5.3

General model for code and text

input
$1.20
output
$2.20
cache
$1.20
glm-5.3-uncensored

Fewer refusals, freer phrasing

input
$2.50
output
$4.50
cache
$0.60
glm-5.3-flash-cybersecurity

Tuned for information security work

input
$0.30
output
$1.00
cache
$0.30
abliterated-large-v2

The largest model in the catalog

input
$6.00
output
$10.00
cache
$0.60

Before a request, a hold is placed on your balance for the maximum possible answer. Afterwards only the real cost is charged and the rest is released.

Top-up

Any amount from the list, paid in crypto through Crypto Bot. The balance is credited automatically.

$1$5$10$25$50$100

FAQ

Which clients work?

Any client that lets you set a custom Base URL for an OpenAI-compatible API: Cline, Roo Code, Kilo Code, Cursor, Continue, Zed, Aider, OpenCode and more. The Anthropic Messages and Responses protocols are not supported.

How many requests can I run at once?

Up to three concurrent requests per key. Above that you get a 429 error; retry in a few seconds.

Why is the answer cut off, or why does the model think for long?

Reasoning models spend part of the tokens on thinking. Set a large output limit in your client, for example 16384.

Do you store my prompts?

We do not store prompt or answer text. We keep service data only: Telegram ID, balance and token accounting.

What if my balance runs out?

The service returns a 402 error. Top up in the bot; you do not need a new key.