DarkGPT FRTry it
API

DarkGPT API: An OpenAI-Compatible AI API, Pay Per Token

Want to plug an AI that doesn’t refuse half your users’ requests into your Discord bot, your website or your scripts? The DarkGPT API gives you a key, a standard-format endpoint and pay-per-token billing, no subscription, with Claude, Gemini, Grok, DeepSeek and the Unfiltered mode models. Here’s everything you need to send your first request.

Updated October 13, 2026By the DarkGPT team
Contents9 sections
  1. Why use the DarkGPT API?
  2. Get started in 3 steps
  3. Authentication
  4. Send a message: /v1/chat/completions
  5. API pricing
  6. An API that doesn’t break your bot with refusals
  7. Rate limits
  8. Error codes
  9. What you can build

Why use the DarkGPT API?

  • Standard format: the /v1/chat/completions endpoint follows the most widely used format, compatible with most existing libraries.
  • Many providers, one bill: Claude, Gemini, Grok, DeepSeek, Mistral, GLM… without opening an account with each one.
  • Pay as you go: billed per 1,000 tokens, from your API & IDE credit, which you can top up from €10.
  • Revoke your key at any time from your “API Keys” page.
Request blocked · content filter enabled
Grok 4.2 Non ReasoningUnfiltered modeper token

My Discord bot has to answer straight, no disclaimers.

Got it. Your bot will answer frankly, no apology paragraph. Here’s the system message to set up.

Get started in 3 steps

  1. Create your keyOn your “API Keys” page on darkgpt.chat. It always starts with dgpt_live_.
  2. Pick your modelEach key is tied to the model you select in your dashboard (for example claude-4.6-sonnet, gemini-3.5, grok-4.2, deepseek-v4-pro, ministral-14b).
  3. Send your requestA simple POST with your messages, and the response comes back as JSON.

You’ll need credit: buy API & IDE Credit on the shop (from €10), then send your order number and your account email to support so it gets added.

Authentication

The API uses Bearer keys. The base URL is https://darkgpt.chat/v1 and every request carries this header:

Authorization: Bearer dgpt_live_xxxxxxxxxxxxxxxxxxxx

Security: keep your key secret. Never put it in a public repo or in code that runs in the browser: call the API from your server. If it leaks, revoke it and create a new one.

Send a message: /v1/chat/completions

JavaScript

const response = await fetch("https://darkgpt.chat/v1/chat/completions", {
  method: "POST",
  headers: {
    "Content-Type": "application/json",
    "Authorization": "Bearer dgpt_live_YOUR_API_KEY"
  },
  body: JSON.stringify({
    model: "claude-4.6-sonnet",
    messages: [
      { role: "system", content: "You are a direct, concise assistant." },
      { role: "user", content: "Explain reverse engineering to me." }
    ],
    temperature: 0.7
  })
});
const data = await response.json();
console.log(data.choices[0].message.content);

Python

import requests

res = requests.post(
    "https://darkgpt.chat/v1/chat/completions",
    headers={"Authorization": "Bearer dgpt_live_YOUR_API_KEY"},
    json={
        "model": "claude-4.6-sonnet",
        "messages": [{"role": "user", "content": "Summarize this text in 3 bullet points."}],
    },
)
print(res.json()["choices"][0]["message"]["content"])

cURL

curl -X POST https://darkgpt.chat/v1/chat/completions \
  -H "Authorization: Bearer dgpt_live_YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"claude-4.6-sonnet","messages":[{"role":"user","content":"Hello"}]}'

Parameters

ParameterTypeWhat it does
modelstring, requiredThe model ID, for example claude-4.6-sonnet
messagesarray, requiredThe { role, content } messages: system, user or assistant
temperaturenumber, optionalCreativity, from 0 to 2 (default 1)
max_tokensinteger, optionalMaximum number of tokens generated
streamboolean, optionalLeave it on false for now: external streaming is coming soon

The response holds the text in choices[0].message.content and a usage block with the token count and the request cost in euros (cost_eur). To list the available models: GET /v1/models.

API pricing

Billing is per 1,000 tokens, with one price for the tokens you send (input) and one for the tokens generated (output). Reminder: 1,000 tokens ≈ 750 words.

ModelInput / 1,000 tokensOutput / 1,000 tokens
Mistral 3B€0.00075€0.00225
Mistral 12B€0.00113€0.00338
Mistral 14B€0.00150€0.00450
Grok 3 Mini€0.00225€0.00675
DeepSeek V4 Flash€0.00281€0.00844
DeepSeek V4 Pro€0.00469€0.01406
Gemini 3.5€0.00750€0.02250
GLM 5.2€0.00844€0.02625
Grok 4.2€0.00900€0.02800
Claude Haiku 4.5€0.00938€0.03000
Claude 4.5 Sonnet€0.01500€0.04688
Claude 4.6 Sonnet€0.01688€0.05250

Example: with Claude 4.6 Sonnet, a request with 400 input tokens and 600 output tokens costs about €0.0383. The simulator at darkgpt.chat/api works out your daily and monthly budget.

An API that doesn’t break your bot with refusals

The worst thing for a bot or an app is a refusal in the middle of a normal conversation: the user asks an adult question and the API returns a generic apology. Your product looks crippled, and it isn’t even your fault.

With the DarkGPT API, you pick an Unfiltered mode model (Grok, DeepSeek, Qwen, Mistral, Kimi…) and set the tone in your system message: frank, funny, blunt, no disclaimers. Your bot answers the way you decided.

Good habit: a small model for volume (Mistral 14B, Qwen 3 235B), a bigger one for the answers that matter. Adults 18+ only, and illegal use remains banned.

Rate limits

API planRequests / minTokens / min
Standard6040 000
Premium300200 000

Limits apply per key. The X-RateLimit-Remaining and X-RateLimit-Reset headers let you track your usage in real time.

Error codes

CodeMeaningWhat to do
401Invalid or missing keyCheck the Authorization header
403Access deniedKey revoked: create a new key
429Too many requestsLimit reached: wait a few seconds
500Server errorRetry; if it keeps happening, contact support

What you can build

  • A Discord bot: one !ask command and your server has its own assistant, running the model of your choice.
  • An assistant on your website: customer support, a smart FAQ or a built-in chat, called from your backend.
  • Your automations: summaries, translations, email sorting or content generation in your scripts, n8n or your own tools.

Frequently asked questions

Is the DarkGPT API OpenAI-compatible?

It uses the standard /v1/chat/completions format (messages, model, temperature, max_tokens), compatible with most existing libraries. Just change the base URL and the key.

Is the DarkGPT API uncensored?

The Unfiltered mode models answer legitimate adult requests without pointless refusals or sermons, and you set the tone with your system message. Illegal use remains banned by the terms.

How much does the API cost?

It’s billed per token, from €0.00075 per 1,000 input tokens with Mistral 3B. API & IDE credit can be topped up from €10, no subscription.

Do I need a DarkGPT subscription for the API?

No. The API uses API & IDE credit, which is separate from chat subscriptions.

What are the rate limits?

60 requests and 40,000 tokens per minute on Standard, 300 requests and 200,000 tokens per minute on Premium, per key.

Is streaming available?

Not yet for the external API: leave stream on false. It’s coming soon.

Try it free