Dikaya

About Dikaya

AI for code and writing. Browser chat, IDE connections and an API — with one prepaid balance.

One service, different ways to work

Use the browser for a conversation, a compatible IDE for project work, or your own application for repeated requests. Your personal Dikaya key connects supported clients to the same account balance.

Client tools, project files and internet access depend on the application and the permissions you grant it. The browser chat does not provide native web search or run commands. Protocol compatibility does not mean every feature of every editor is supported.

Models, with their actual names

The five new GLM, Qwen and DeepSeek variants are available alongside Dikaya Flash, Pro and Pro 2. Modified versions are labelled as such; DeepSeek 1017 Safeguards is a separate original variant. Your authenticated /v1/models catalogue is authoritative for permitted models and personal rates.

Dikaya Flash

seller-flash

Quick questions, short edits and direct image understanding.

Context: 256,000 tokens. Output and request limits are separate.

Model details

Dikaya Pro

seller-pro

Large-context analysis, writing and long conversations.

Context: 1,000,000 tokens. Output and request limits are separate.

Model details

Dikaya Pro 2

seller-pro-2

A modified version based on GLM-5.3. Code, complex tasks and creative writing. Up to 1 million tokens of context. The default demo model.

Context: 1,000,000 tokens. Output and request limits are separate.

Model details

GLM 5.3 Abliterated

seller-glm-5-3

Modified abliterated model. Text input, low/high/max reasoning. Paid access.

Context: 1,048,576 tokens. Output and request limits are separate.

Practical guide (Russian)

GLM 5.3 Flash Abliterated

seller-glm-5-3-flash

Modified abliterated model. Text input, low/high/max reasoning. Paid access.

Context: 1,048,576 tokens. Output and request limits are separate.

Practical guide (Russian)

Qwen3.8 Flash Next Abliterated

seller-qwen-3-8-flash-next

Modified abliterated model. Text input, low/high/max reasoning. Paid access.

Context: 262,144 tokens. Output and request limits are separate.

Practical guide (Russian)

DeepSeek V4.1 Flash Abliterated

seller-deepseek-v4-1-flash

Modified abliterated model. Text input, low/high/max reasoning. Paid access.

Context: 1,048,576 tokens. Output and request limits are separate.

Practical guide (Russian)

DeepSeek V4.1 Flash 1017 (Safeguards)

seller-deepseek-v4-1-flash-1017

Original Safeguards, not abliterated

Context: 1,048,576 tokens. Output and request limits are separate.

Practical guide (Russian)

Top up when you need to

Chat and API use prepaid credit, without a subscription or balance expiry. Top-ups start at $5. Package bonuses are added only after a confirmed qualifying payment; creating an invoice does not credit the account.

Usage is charged by ordinary input, cached input and output tokens. Reasoning is included in output, not billed twice. The service reserves an estimate before processing and settles actual usage when it is known. Image description, if used, is a separately billed step.

Current rates and top-up packages

Start with your own task

Eligible new accounts receive up to 3 requests on Pro 2. Demo is text-only, with up to 8192 output tokens and lower input limits. Earlier grants keep their saved conditions. Browser/network eligibility controls apply; creating an account does not guarantee a new demo grant.

Choose a small coding, editing or analysis task that you can check yourself. Generated output can be incomplete or inaccurate. The model name and context window are not a benchmark or a guarantee of an answer to every request.

Open your account and chat

Data and support

Data processing, access and deletion limits are described in the privacy notice. Payment and technical issues are reviewed through Support inside your account; no response-time SLA or automatic refund is promised.

Privacy notice · Contact and support

Sources and editorial information

Dikaya publishes these pages about its own service. Product facts are checked against the current model catalogue, pricing and implemented account behavior. Practical guides may be AI-assisted; examples are illustrative unless explicitly identified as measured results. No independent benchmark or search-ranking guarantee is claimed.

Practical guides (Russian)

Reviewed