What is Unrestricted?

A chat interface to frontier-class large language models, deployed without the consumer-product moderation stack that causes most refusals elsewhere.

Is Unrestricted a jailbreak?

No. It's a different deployment. Nothing is being bypassed at inference time — the moderation stack simply isn't shipped.

What can I ask about?

Medicine, law, history, chemistry, politics, security research, fiction of any genre, philosophy, religion, or anything else you'd expect a good reference book to cover.

What can't I ask about?

A narrow floor: facilitating violence against specific named people, sexual content involving minors, authoring live weaponized malware for identified targets. That list is short and doesn't move.

Do you store my conversations?

Yes — to your own account, so you can return to old threads. We don't use your chats for training, and we don't share them with third parties. You can delete any thread from your history at any time.

Do you train on my chats?

No. Your conversations never enter any training pipeline.

Do you log my prompts?

Your conversations sit in your own account history so you can return to them. We don't aggregate prompts into a training corpus, we don't sell them, and we don't share them with third parties. Rate-limiting and abuse detection run on per-session counters.

Which model is under the hood?

A frontier-class model. We don't publish the exact version because it moves; we do publish benchmark results so you can see how it performs.

Is this ChatGPT with the filters off?

No. It's not ChatGPT at all. The base model is different, the deployment is ours, and the lack of moderation stack is a deliberate architectural choice.

Yes. We comply with applicable law, including content that's actually illegal. The category of things we decline is narrower than the category other products decline — not wider than the law.

Can I use Unrestricted for professional work?

Yes. Researchers, clinicians, security professionals, writers, and lawyers are common users. Use of any AI output is still your professional judgement.

API access is coming. Waitlist on the homepage.

How much does it cost?

There's a free tier. Paid plans add usage headroom, not different moderation behavior. The content policy is the same at every tier.

Will Unrestricted write erotica?

Yes, between consenting adults. We don't write CSAM under any framing.

Will it help me with a CTF or a bug-bounty engagement?

Yes. Training labs, public research, published CVEs, decompilation help — all fair territory.

Will it help me with homework?

It will help you think through the problem, which is the educationally useful thing. Direct 'write my essay' requests get an answer, but it's not the thing that will make you a better writer.

How is this different from running a local open-weights model?

Our model is larger, faster, and better than any open-weights model you can run on consumer hardware. It's also reachable from a browser without a GPU.

Can I use Unrestricted to hurt someone?

No. The floor we keep is precisely about that — we won't help plan or facilitate violence against identified people.

Is Unrestricted political?

We'll discuss politics in any direction, with anyone, from any angle. We don't have a partisan preference and we don't inject one into the answers.

Can I trust the answers?

About as much as any frontier model — which is to say, verify anything load-bearing. We're a powerful reference, not an oracle.

Do you have a mobile app?

The web app works well on mobile. Native apps are on the roadmap.

Who's behind Unrestricted?

A small team with backgrounds in ML, infrastructure, and product. We publish our founding team and investors on the About page.

Why does this blog exist?

To show our work — benchmarks, methodology, product philosophy. It's also the most honest way to explain the mission to people who are skeptical of the pitch.

How do I give feedback?

There's a feedback button in the chat UI. It goes to a human.

What's the long-term plan?

Keep refusals at a narrow floor. Keep privacy as architecture. Ship a product that treats adults like adults. The plan is to stay boring about the boring parts and principled about the principled ones.

back to blog

FAQJune 9, 2026·8 min read

25 questions we keep getting about Unrestricted

What can I ask? What can't I ask? Do you store anything? Is this legal? Is this a jailbreak? The unfiltered FAQ, in one place.

We get the same twenty-five questions on a loop. Here they are, with straightforward answers. If your question isn't here, ask the model — that's kind of the point.

Frequently asked

What is Unrestricted?
A chat interface to frontier-class large language models, deployed without the consumer-product moderation stack that causes most refusals elsewhere.
Is Unrestricted a jailbreak?
No. It's a different deployment. Nothing is being bypassed at inference time — the moderation stack simply isn't shipped.
What can I ask about?
Medicine, law, history, chemistry, politics, security research, fiction of any genre, philosophy, religion, or anything else you'd expect a good reference book to cover.
What can't I ask about?
A narrow floor: facilitating violence against specific named people, sexual content involving minors, authoring live weaponized malware for identified targets. That list is short and doesn't move.
Do you store my conversations?
Yes — to your own account, so you can return to old threads. We don't use your chats for training, and we don't share them with third parties. You can delete any thread from your history at any time.
Do you train on my chats?
No. Your conversations never enter any training pipeline.
Do you log my prompts?
Your conversations sit in your own account history so you can return to them. We don't aggregate prompts into a training corpus, we don't sell them, and we don't share them with third parties. Rate-limiting and abuse detection run on per-session counters.
Which model is under the hood?
A frontier-class model. We don't publish the exact version because it moves; we do publish benchmark results so you can see how it performs.
Is this ChatGPT with the filters off?
No. It's not ChatGPT at all. The base model is different, the deployment is ours, and the lack of moderation stack is a deliberate architectural choice.
Is this legal?
Yes. We comply with applicable law, including content that's actually illegal. The category of things we decline is narrower than the category other products decline — not wider than the law.
Can I use Unrestricted for professional work?
Yes. Researchers, clinicians, security professionals, writers, and lawyers are common users. Use of any AI output is still your professional judgement.
Is there an API?
API access is coming. Waitlist on the homepage.
How much does it cost?
There's a free tier. Paid plans add usage headroom, not different moderation behavior. The content policy is the same at every tier.
Will Unrestricted write erotica?
Yes, between consenting adults. We don't write CSAM under any framing.
Will it help me with a CTF or a bug-bounty engagement?
Yes. Training labs, public research, published CVEs, decompilation help — all fair territory.
Will it help me with homework?
It will help you think through the problem, which is the educationally useful thing. Direct 'write my essay' requests get an answer, but it's not the thing that will make you a better writer.
How is this different from running a local open-weights model?
Our model is larger, faster, and better than any open-weights model you can run on consumer hardware. It's also reachable from a browser without a GPU.
Can I use Unrestricted to hurt someone?
No. The floor we keep is precisely about that — we won't help plan or facilitate violence against identified people.
Is Unrestricted political?
We'll discuss politics in any direction, with anyone, from any angle. We don't have a partisan preference and we don't inject one into the answers.
Can I trust the answers?
About as much as any frontier model — which is to say, verify anything load-bearing. We're a powerful reference, not an oracle.
Do you have a mobile app?
The web app works well on mobile. Native apps are on the roadmap.
Who's behind Unrestricted?
A small team with backgrounds in ML, infrastructure, and product. We publish our founding team and investors on the About page.
Why does this blog exist?
To show our work — benchmarks, methodology, product philosophy. It's also the most honest way to explain the mission to people who are skeptical of the pitch.
How do I give feedback?
There's a feedback button in the chat UI. It goes to a human.
What's the long-term plan?
Keep refusals at a narrow floor. Keep privacy as architecture. Ship a product that treats adults like adults. The plan is to stay boring about the boring parts and principled about the principled ones.

Ready to experience an AI without a leash?

Start chatting free