Skip to content
DikayaChat

AI chat without censorship for coding and creative work

Version reviewed

Bring your own question, scene or code to Dikaya. Work in the browser, compare saved variants and keep control of what you send.

What “without censorship” means here

It describes Dikaya’s positioning around open-ended coding, writing and discussion, not a measured zero-refusal rate. Your task is not confined to a menu of canned scenarios. Responses still depend on the model, context, instructions and technical limits; an answer can be refused, wrong or incomplete.

Start with a small task you can judge

For code, provide inputs, expected output and dependency constraints. For writing, give audience, voice and facts that must survive editing. For fiction, specify adult characters, motives and the change the scene should produce. Ask for alternatives before a long draft, then refine a concrete weakness rather than simply saying “better”. The example below is a task you can edit, not evidence of a model result.

Choose a model, not a promise

Dikaya Pro 2 is a modified version based on GLM-5.3; its exact API ID remains seller-pro-2. This attribution says nothing about other tiers’ foundations. Consult the current model catalogue for available models, context, image handling and prices. A large context window is not a promise of equally long output or low cost.

Try, then decide how to continue

Register with username/password, no email. An eligible new account may receive up to 3 one-time Pro 2 requests: text only, 8192 output tokens including reasoning, 6000 context bytes, 10 requests/minute and one concurrent request. Existing allowances stay intact. Paid output defaults to 16384 unless your key/client specifies otherwise; the explicit gateway maximum is 65536, subject to the model. Shared prepaid credit has no subscription or expiry; files and explicit web search require top-up.

Freedom of topic is not zero retention

Server history, branches and service prompt/answer/usage logs are stored. The administrator has protected access and private Telegram notices; necessary context goes to external model processing. Search sends its visible query to external engines. Deleting chat does not erase every financial, moderation or backup record. Use the API page for application integration, not as a different privacy promise.

Illustrative task, not a captured model answer

Help me outline a short story scene with two adult siblings preparing a family celebration. One wants to preserve tradition; the other wants a change. Give three conflicts with understandable motives and distinct voices. Do not write the ending: I will choose the direction first.

This opens the visible task as an editable chat draft after sign-in. Only your explicit send starts processing; this page generates nothing.

Questions and answers

Does every prompt get an unrestricted answer?

No. This is a service positioning, not a universal answer or accuracy guarantee. Compare models using the same bounded tasks and your own acceptance criteria.

Is there unlimited free chat?

No. The one-time eligible demo is limited; paid chat and API share a prepaid token-based balance without a subscription or balance expiry.

Does Stop make the request free?

No. Completed processing can be charged and an unknown outcome can retain a reservation. Viewing a saved branch is not generation; edit-and-generate or regenerate is a new request.

Sources and service conditions