Unrestricted FAQ: the model, its limits, accounts and plans
Answers about Unrestricted: which model it uses, what it still refuses, accounts on the web and iPhone, what is stored, and how the free allowance and Pro work.
By Unrestricted team · Updated November 2026 · Editorial standards
Unrestricted is a text AI chat on the web. It answers through one uncensored model based on Dolphin-Mistral-24B, a 24B open-weights model, served by a third-party inference provider. We add no filter or classifier of our own on the web, but the provider still refuses some prompts, and our Terms ban some uses.
This page answers questions about the product itself: the model, its limits, accounts, storage and plans. For what "no restrictions" can mean across the whole category, and the limits every chatbot still has, see AI without restrictions: what it means, and its limits.
Which model answers
On the web, Unrestricted answers through one uncensored model based on Dolphin-Mistral-24B, a 24B open-weights model. A third-party inference provider serves it, we didn't train it, and we add no filter or classifier on top of it. Why we chose that setup is in why we built Unrestricted.
Unrestricted isn't a jailbreak. A jailbreak tries to talk a model out of its own rules; Unrestricted uses a model that is uncensored to begin with. Why jailbreaks keep breaking is covered in ChatGPT DAN and other jailbreaks.
Open-weights Dolphin models can also run locally. The Hugging Face card for Dolphin 3.0 Mistral 24B, a separate Dolphin release built on Mistral Small 24B (not the one Unrestricted is based on), lists Ollama and LM Studio among the ways to run it. [1]Unrestricted gives you a hosted model based on Dolphin-Mistral-24B, served by a third-party inference provider, on the web, so there's nothing to download. Where the Dolphin models come from and how they're trained is in our Dolphin Mistral explainer.
What it still refuses
On the web we add no filter or classifier of our own, but that doesn't mean every prompt gets an answer. The inference provider applies its own content policy and still refuses some prompts.
Our Terms also ban some uses. Section 4 says you may not use Unrestricted for any of the following. [2]
- 01
Violence against a person
Planning, facilitating or carrying out violence against an identified person.
- 02
Sexual content involving minors or real people
Sexual content involving minors, or content that sexualizes real, identifiable people without their consent.
- 03
Targeted malware
Functional malware, ransomware or exploit code aimed at a specific victim, system or organization without clear authorization.
- 04
Impersonation
Content designed to impersonate a specific real person for fraud, harassment or non-consensual depiction.
- 05
Breaking the law
Violating the law, infringing other people's rights, or evading sanctions or export controls.
- 06
Attacking the service
Attacks on Unrestricted itself, such as denial of service, scraping at unreasonable volume, or probing for vulnerabilities outside an authorized research program.
Answers can also be wrong. The Terms say outputs can be wrong, outdated, biased or fabricated, and that you shouldn't base medical, legal, financial or other consequential decisions on them without checking a qualified source. [2]
Accounts, platforms and subscriptions
Unrestricted runs in a web browser at unrestricted.chat. Our iPhone app (iOS 18 or later) isn't available on the App Store right now. There's no Android, desktop or native iPad app, and there's no public API.
The web needs a Google or verified-email account. The iPhone app doesn't ask for one.
Web and iPhone subscriptions are billed separately. How web plans are billed and cancelled is in section 8 of our Terms. [2]
Storage, incognito and training
When you send a message, your prompt goes to a third-party inference provider to generate the reply. Saved chats are stored on our servers.
On the web, an incognito chat isn't saved to our database. It has to start as a new chat, and your prompt, along with any image you attach, still goes to the model provider.
We don't train models on your chats, and we don't sell them. [2] How that compares with the training defaults at OpenAI, Anthropic and Google is in our comparison of how ChatGPT, Claude and Gemini train on chats.
The free allowance and Pro
There's a free allowance to try it, then Pro. On the web the free allowance is one-time; it doesn't reset.
On the web, Pro removes the message quota. Per-minute rate limits, counted per user, still apply. iPhone limits are set separately.
Web search with citations is included at no extra cost; it isn't a Pro feature. You can also attach an image and ask about it.
Support
Email hello@unrestricted.chat, or text +1 (363) 333-0075. Both are listed on our public support page.
Last reviewed November 2026. If an answer here is wrong or missing, tell us at hello@unrestricted.chat. And if you're still comparing options, start with what "AI without restrictions" can and can't mean.
Frequently asked
What is Unrestricted?
Unrestricted is a text AI chat on the web. It answers through one uncensored model based on Dolphin-Mistral-24B, a 24B open-weights model, served by a third-party inference provider. We add no filter or classifier of our own on the web, but the provider still refuses some prompts, and our Terms ban some uses.
Which model does Unrestricted use?
On the web, Unrestricted answers through one uncensored model based on Dolphin-Mistral-24B, a 24B open-weights model. A third-party inference provider serves it, and we didn't train it. We add no filter or classifier of our own on the web.
What does Unrestricted still refuse?
On the web we add no filter or classifier of our own, but the inference provider applies its own content policy and still refuses some prompts. Our Terms also ban some uses: violence against an identified person, sexual content involving minors or sexualizing real people without consent, malware or exploit code aimed at a specific target without authorization, impersonating a real person for fraud or harassment, breaking the law or evading sanctions, and attacking the service.
Do I need an account?
On the web, yes: it needs a Google or verified-email account. The iPhone app doesn't ask for one, but it isn't available on the App Store right now. Web and iPhone subscriptions are billed separately.
Is there an iPhone app?
Not right now. Our iPhone app (iOS 18 or later) isn't available on the App Store at the moment, and there's no Android, desktop or native iPad app. If you already have the iPhone app, its limits are set separately from the web, and the two subscriptions are billed separately.
Sources
Checked November 2026.
- Dolphin 3.0 Mistral 24B model card, dphn on Hugging Face
- Terms of Service, sections 4 (Acceptable use), 5 (Your content), 6 (Model outputs) and 8 (Payments), Unrestricted
Keep reading
AI limits: message caps and rate limits in ChatGPT, Claude and Gemini
How AI usage limits work: message caps, rate limits and context windows in ChatGPT, Claude and Gemini, what 'unlimited' means, and how to tell which you hit.
ChatGPT Temporary Chat: what it saves, what it skips, and how long OpenAI keeps it
ChatGPT Temporary Chat skips your history, memory and training, but OpenAI can keep a copy for a while. How to start one, and how Claude and Gemini compare.
Dolphin Mistral: what the Dolphin models are and how they're trained
What Dolphin Mistral is, who makes the Dolphin models, how they are fine-tuned to follow the system prompt instead of refusing, and their licenses.