~/minaka/new conversation
日本語
MINAKA CONSOLEBYOK / v0.2

A place to think.

Write a story. Follow a question. Give an idea a shape.

Bring your own API key. Talk to a model you choose.

› Start with your API key, model and consent in connection settings

Requests include conversation history. Clear long chats. HTML and Markdown are displayed as text, never executed.

Connection & model

1. Set up your key and model

Create an OpenRouter API key and purchase credits if needed. Set a spending limit on that key at the provider. Free models also have usage limits.

API keys & spending limits ↗ · Provider credits & billing ↗

Used only in this tab, without storage or logging. Avoid shared devices. Never paste a key into the chat message.

MODEL RANKINGS

Choose a model from the rankings

Choosing a model does not run AI. Prices refresh when models load. Weekly usage uses the key above for OpenRouter authentication.

Input, history and reasoning may also be billed. This token limit is not a monetary cap. Your provider billing records are authoritative.

2. Review the terms

AI chat terms & privacy (2026-10-25 / v2)
  1. Operated by Studio Toriumi, info@toriumis.com. This UI is free. AI service, charges, credits and refunds follow your agreement with OpenRouter. No payment contract with this site is created.
  2. On send, your key goes to OpenRouter for authentication. Prompts and conversation history go to OpenRouter and the selected model provider. Review their retention and training settings. Do not enter personal, confidential or third-party private information.
  3. Keys and chats are not sent to this site's server or stored in localStorage, cookies or a database. They remain in the page and tab memory until reload, close or disconnect. This does not delete data already sent to providers. Hosting may process ordinary access information.
  4. Each send may incur provider charges. Requests are not automatically retried. Stopping or timing out may still leave billable usage; check provider records before retrying.
  5. Illegal use, infringement, harassment and disruption are prohibited. Provider and model policies apply. Omniscience, uncensored output, accuracy and profit are not guaranteed. Verify outputs before important decisions.
  6. The service may change or stop. Liability follows applicable law; these terms do not waive all liability or liability for intentional misconduct or gross negligence. Japanese law applies without restricting mandatory rights. Review revised terms before use. Refer to Japanese if translations differ.

OpenRouter Terms · OpenRouter Privacy

FIELD NOTES

API FIELD NOTES / 2026-10-25

Choosing a less restricted AI

Criteria, in order: compatibility with this UI, documented model policy, then pricing transparency. These picks use official specifications (evidence tier 1). We have not benchmarked paid responses, Japanese quality or refusal rates. These links carry no referral commission.

01 / START WITH THIS UI

OpenRouter → Venice: Uncensored

A candidate available with your existing OpenRouter key. Look for “Venice: Uncensored” in the model list. No separate provider key is needed.

cognitivecomputations/dolphin-mistral-24b-venice-edition

$0.20 input / $0.90 output per million tokens

Tradeoff: review routing and provider data policies. An “Uncensored” model name does not remove provider terms or all model refusals.

Official model page ↗
02 / DIRECT SPECIALIST API

Venice API

A candidate if you want a direct relationship with a service that markets uncensored AI. Its official pricing lists Venice Uncensored 1.2 and Role Play Uncensored.

Uncensored 1.2: $0.20 / $0.90
Role Play: $0.50 / $2.00 USD / 1M input / output tokens

Tradeoff: direct Venice keys do not work in this UI; use another client or add an integration. Data handling differs by model, so review the official table.

Official API pricing ↗ · Service positioning ↗
03 / COMPARE MORE OPTIONS

NanoGPT

A candidate for comparing multiple uncensored-oriented models. It offers an OpenAI-compatible API and lists Uncensored and Abliterated models on its official pricing page.

Pay-as-you-go available; check each model’s price

Tradeoff: NanoGPT keys do not work in this UI. Verify model IDs, reasoning options and billing despite API compatibility. Its commercial models are not all uncensored.

Official pricing ↗ · API docs ↗

Prices are snapshots checked on October 25, 2026. Review current prices, terms, retention and training settings before purchasing. “Uncensored” is provider terminology, not permission for illegal use or a guarantee of unlimited access, confidentiality or zero refusals.