This list of uncensored AI tools is arranged by the job you want done rather than by brand: browser chat, roleplay, the API, images, video, coding, scenario writing, local weights, and two hosted apps described from their own pages. Eight of the ten run on one prepaid key here, and their prices are read from the live catalog every time the page renders. Catalog as of 2026-11-16. To try a model before anything else, /chat gives three free turns a day with no account.
Ten tools, one per job
Most lists ranking for this query sort by brand name. This one sorts by job, since one model does different work in each tool. Eight of the ten run on a single prepaid key on this site. The last two are hosted apps that also rank for the query, described from their own pages as read on October 27, 2026.
| # | Job | Tool | What it gives you |
|---|---|---|---|
| 1 | Talking to a model in the browser | /chat, plus the playground | Three free turns a day with no account; the playground runs every per-token build once you paste a key |
| 2 | Character roleplay | Roleplay page, or SillyTavern | Cards you can export as SillyTavern JSON, and one model id that works in SillyTavern or Janitor AI |
| 3 | An app or agent you write | The API | One base URL, one key, the OpenAI chat-completions format, billed per token |
| 4 | Making images | Image models | 10 uncensored image models, billed per image |
| 5 | Making video | Video models | 3 uncensored video models, billed per second |
| 6 | Coding, including security work | Code builds inside Claude Code or Cursor | How to wire the uncensored code builds into an agent; the security builds have their own page, red teaming |
| 7 | Writing scenarios | Scenario generator | A roleplay scenario written by an uncensored model, with two free runs |
| 8 | Weights on your own GPU | Ollama or LM Studio | The same builds as GGUF downloads, run on your own GPU with no per-token bill |
| 9 | Private chat as a hosted app | Venice | Free plan with daily prompt caps, paid monthly plans (read October 27, 2026) |
| 10 | Any Hugging Face model on a flat plan | Featherless | Flat monthly plans across a large open-model catalog (read October 27, 2026) |
On the chat builds, "uncensored" has one meaning: the refusal has been removed from the weights, either by abliteration or by an uncensored fine-tune. The two edits are explained on abliterated vs uncensored.
The price of the builds behind them
Three builds, one of each kind, with prices read from the catalog as this page rendered: the assistant at the top of the models page, one roleplay tune, one security build. These are the per-token rates every tool on the key is billed at.
| GLM-5.3 Uncensored | Uncensored | $0.282 | $0.846 |
| Cydonia 24B v4.3 | Roleplay | $0.484 | $0.60 |
| GLM-5.3 Flash Cybersecurity | Security | $0.5093 | $1.6975 |
Once a day /proof checks that the id you send is the build its label claims. On November 16, 2026, 146 of the 168 listed chat models that were checked came back stable. Top up from $10 by card, WeChat Pay or crypto; usage settles per token.
1. Browser chat: /chat and the playground
/chat runs the builds in a browser tab with nothing to install: three free turns a day, no account, no email, no phone. System prompt and temperature sit in Settings, and chats stay in the browser. The playground is the developer’s version of the same thing, with a model picker over every per-token build, so a question one build hedges on moves to another in a click; its opening turns need no key on the lower-priced builds. When the chatbot is only a means and the model is the point, this is what people searching "best AI chatbot with no restrictions" want.
2. Roleplay: cards in the browser, or SillyTavern
Character cards sit on top of the same chat on the roleplay page, and each card exports as SillyTavern JSON, so a character made in the browser loads into the desktop client as is. SillyTavern and Janitor AI take three values in their proxy settings: base URL, key and model id; the docs page shows where they go for SillyTavern and for Janitor AI. Roleplay tunes answer straight away with no reasoning step, which suits a card client.
3. The API: every build behind one key
For code of your own. Whatever the language, the OpenAI SDK needs two values swapped:
- Base URL
- https://api.unrestrictable.ai/v1
- Key
- sk-unr-live-…Made in the console and shown one time
- Model
- glm-5.3-uncensoredAny id on the models page
Each response reports the settled cost of that request in a header. There are guides for Open WebUI, LiteLLM, n8n and a Discord bot under guides; the docs page carries Python and Node samples.
4 and 5. Images and video: the media models
Media works the same way: open image and video models on the same key, billed per image or per second, with a media console to run them in. Two rules apply to each media prompt: no depiction of a minor, and no editing a real person’s photograph without that person’s consent; the acceptable use policy has the wording. The media catalog lists the tiers and how to format a request.
| Per | Endpoints | |||
|---|---|---|---|---|
| LTX-2.3 Image-to-Video Uncensored | Video | $0.08 | second | |
| Wan 3.0 Image-to-Video Uncensored | Video | $0.20 | second | |
| Wan 3.0 Video Uncensored | Video | $0.20 | second | |
| Anime Image Uncensored | Images | $0.03 | image | |
| Chroma Uncensored | Images | $0.08 | image | |
| CyberRealistic XL Uncensored | Images | $0.0136 | image | |
| Qwen Edit Uncensored | Images | $0.08 | image | |
| Qwen Image 2.1 Uncensored | Images | $0.0375 | image | |
| Qwen Image 3 Uncensored | Images | $0.08 | image | |
| Seedream 5.0 Lite Edit Uncensored | Images | $0.0313 | image | |
| Vision Image Uncensored | Images | $0.04 | image | |
| Vision Pro Image Uncensored | Images | $0.09 | image | |
| WAI Illustrious SDXL Uncensored | Images | $0.0136 | image |
6. Code and security: the code builds inside an agent
The code builds (uncensored-coding-agent, glm-5.3-flash-code-uncensored) treat exploit, malware-analysis and payload questions as code to read, not a sermon. They run in Claude Code through a router, in Cursor’s model settings and in Open WebUI; the uncensored coding AI guide sets each up and prices a day of coding at current rates. The 7 security builds and the agent settings are on red teaming. Testing a system you do not own, or lack written permission to test, stays forbidden under acceptable use.
7. Scenarios: the AI scenario generator
Give the AI scenario generator a genre and a few constraints and an uncensored model writes a roleplay scenario, with two runs free. It is the quickest way from "a heist in a drowned city" to a loadable card; the long version is roleplay ideas by genre.
8. Local: the same weights in Ollama or LM Studio
Most builds on the key have public weights with GGUF files on Hugging Face, so searches for "uncensored AI github" and "run it locally" land on the same models. Running them yourself means a GPU, no per-token bill, and prompts that never leave the machine. The Ollama guide handles the download, LM Studio opens the same GGUF files, and the family pages link each build’s Hugging Face card, one click away from its API page.
9 and 10. Venice and Featherless, in their own words
Venice is a private chat app with an API attached. On October 27, 2026 its pricing page showed a Free plan at $0 with base models, 10 text prompts and 15 image prompts a day; Pro at $18 a month with unlimited text prompts, 1,000 images a day and 100 credits a month; and Pro Plus ($68) and Max ($200) adding API spend. Where it wins is private chat with nothing to set up. The vendor picks the model, and you pay by the month. The Venice comparison prices the same month on both sides.
Featherless serves open models from Hugging Face on flat plans. Its pricing page on October 27, 2026 showed Chat at $25 a month with 4 concurrent units, Developer at $50 a month in credits with 100 concurrent units, and a Business tier, over a catalog it counted at 40,000+ open models. Where it wins is range. You pay by plan and concurrency, not per token.
Is uncensored AI safe to use?
That is two questions. Safe for you: by API the request body is not retained unless a key opts into the 7-day replay, there is no chat history on a server to browse, and every key carries its own spend limit. Safe for other people: the model writes what it is asked, so the limits live in the acceptable use policy: no minors in any form, no real people without their consent, no attacks on systems that are not yours. On every image and video model, one check inside the API rejects a prompt describing a minor before it is priced.
A build with its refusal direction removed still declines now and then; the daily check on the proof page shows how often. Keys come from the console, prepaid from $10, no subscription needed.
Questions
Pick by job: /chat or the playground for conversation, the roleplay page or SillyTavern for characters, the API for your own software, the media models for image and video, the code builds in Claude Code or Cursor for programming, and Ollama or LM Studio to run weights locally. Every one uses builds with the refusal removed from the weights; 169 are listed today. See every uncensored model with its price.
The weights sit on Hugging Face. What GitHub holds is the tooling (Ollama, llama.cpp, Open WebUI, SillyTavern) and the abliteration code itself. Each build’s Hugging Face card is linked from its family page here, and the same model ids answer through the API without downloading anything. See the models by family.
The model answers whatever you ask, which moves the rules into the acceptable use policy: it bars anything involving a minor, any real person without consent, and attacks on systems you do not own. Every image and video model runs one check in the API that refuses a prompt describing a minor before pricing. Unless a key opts into replay, request bodies are not retained. See read the acceptable use rules.
Yes. The code builds return exploit and malware-analysis answers as code, inside Claude Code, Cursor or Open WebUI. Setup is in the coding guide; the security builds are on the red-teaming page. See how to set up uncensored coding AI.
glm-5.3-uncensored heads the models page and is the playground's default general assistant; cydonia-24b-v4.3 is a roleplay tune. Every build's price, context window and daily verdict are on the models page. See what AI with no restrictions means.
Venice describes itself as private, uncensored chat with no card needed on the free plan (read October 27, 2026). As with any hosted app, the model and the filter are the vendor's call; through the API those calls are yours. See Venice side by side.
/chat gives three free turns a day with no account, and the Telegram bot answers three free replies a day. The API needs one key, created once; nothing logs in after that. See the browser chat.
Every model above is on the uncensored model list with its live price per million tokens. Per-token pricing sets the whole catalog side by side, and the family directory groups the builds by base model.