Uncensored AI.

Think bolder. Create faster.
GLM 5.3, Qwen3.8 and DeepSeek V4.1 — modified variants and a separate original Safeguards model, in chat, IDEs and through the API.

Try Dikaya

Try the demo: 3 requests. One grant, subject to demo eligibility limits.

GLM-5.3

Illustration, not a real model response. Prompt: Reproduce CVE-2024-3094 in our sandboxed environment. Illustrative step: read main.go. Example answer: I will reproduce CVE-2024-3094 in our sandboxed environment. Isolating the binary, running the fixture, and capturing the crash — the host stays untouched.

import os
from openai import OpenAI

client = OpenAI(base_url="https://dikaya.cc/v1", api_key=os.environ["DIKAYA_API_KEY"])
reply = client.chat.completions.create(
    model="seller-pro-2", max_tokens=512,
    messages=[{"role": "user", "content": "Help me design a small service"}],
)
print(reply.choices[0].message.content)

curl https://dikaya.cc/v1/chat/completions \
  -H "Authorization: Bearer $DIKAYA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "seller-pro-2",
    "max_tokens": 512,
    "messages": [{"role": "user", "content": "Help me design a small service"}]
  }'

Illustration, not a real model response. No requests are sent.Connect the API

Start with your own task

Five new models. Your choice.

Full catalogue

Four modified abliterated variants and a separate original Safeguards model. Exact names, context windows and retail USD rates per 1M tokens are listed below. New models require paid access; the existing Pro 2 demo remains available.

GLM 5.3 Abliterated`seller-glm-5-3`

Modified GLM 5.3. For tasks where you want to compare a larger model on your own code and text.Context: 1,048,576 tokens

Input $4Output $6Cached $0.7

GLM 5.3 Flash Abliterated`seller-glm-5-3-flash`

Modified GLM 5.3 Flash. Lower token rates for repeated edits and short iterations.Context: 1,048,576 tokens

Input $0.5Output $1.6Cached $0.3

Qwen3.8 Flash Next Abliterated`seller-qwen-3-8-flash-next`

Modified Qwen3.8 Flash Next. A 262,144-token window, with output and request limits specified separately.Context: 262,144 tokens

Input $0.36Output $1.2Cached $0.18

DeepSeek V4.1 Flash Abliterated`seller-deepseek-v4-1-flash`

Modified DeepSeek V4.1 Flash. A separate choice for comparing your API workflows.Context: 1,048,576 tokens

Input $1.2Output $3.2Cached $0.3

DeepSeek V4.1 Flash 1017 (Safeguards)`seller-deepseek-v4-1-flash-1017`

Original Safeguards variant, not abliterated. The lowest published token rates in this five-model lineup.Context: 1,048,576 tokens

Input $0.1Output $0.3Cached $0.008

These are ordinary retail prices. Your account and authenticated /v1/models show any personal budget or wholesale tariff. Reasoning uses the output budget; image description, when requested, is a separately billed Flash stage. No subscription or balance expiry.

Ideas without censorship.

Browser chat saves conversations. Your personal API key connects Dikaya to an editor or application. Both share one balance.

OpenCode, Cursor, Continue, Cline and SDK setup is in the reference. The illustration above does not execute code or send a model request.

Open documentation

The account and chat interface currently use Russian.

Top up when needed.

One-time top-ups. No subscription. Your balance does not expire.

Start

$5 creditA simple start without commitment

Choose Start

Builder

$20

$21 credit+5% top-up bonus

Choose Builder

Studio

$50

$55 credit+10% top-up bonus

Choose Studio

Scale

$100

$115 credit+15% top-up bonus

Choose Scale

Custom top-ups start at $5. Bonuses depend on one confirmed payment; deposits are not combined to reach a higher tier.