Free Trial and FAQ: Try the Uncensored Chatbot API

About the free trial

What do I get when I sign up?

Every new account starts with $0.50 of credit. It is valid for 7 days and you are never asked for payment details to get it.

How do I start?

Open the get-your-key page, enter an email and a password, and your API key is displayed right away. There is no waiting for approval and no email round trip before you can make a call.

What can $0.50 actually buy?

Plenty for a prototype. As an assumption, say a chat turn uses 1,000 prompt tokens and 200 completion tokens. At $0.25 per million input and $1.00 per million output, that turn costs $0.00045, so $0.50 covers about 1,100 such turns. Your own prompts will differ, so check the usage figure.

Is the trial limited to certain features?

No. Streaming, tools and every sampling setting work during the trial, and the limits are the same as for any account: 100,000-token context, 300 requests per minute and an 8 MB body.

When the trial ends

What happens after 7 days?

Unused trial credit expires. Requests then return a 402 with the code no_credit until you top up the prepaid balance.

Do I have to subscribe to continue?

No subscription exists. You add prepaid balance when you want to, it is spent per token, and money you add never expires.

Will my key change when I start paying?

Nothing about your account needs re-creating. You keep working with your existing key unless you choose to regenerate it.

How do I know the trial is about to end?

Count seven days from sign-up and keep your own reminder. Your app should also treat a 402 as the signal to show a paused state, then resume once the balance is topped up.

Can I move to production without changing code?

Yes. The base URL, model id and request format stay the same, so only your balance changes. Add your age gate, limits and error handling before real users arrive.

Can I lose my balance by being inactive?

No. Prepaid balance never expires, so a quiet month does not shrink it. Only the free trial credit has the 7-day window.

Stretching the trial credit

Seven days and fifty cents go a long way when you are deliberate. The table shows what the credit covers for three imaginary chat shapes. Every token count is an assumption chosen to make the arithmetic easy to follow; measure your own with the usage field before planning around these numbers.

Scenario (assumed tokens per turn)	Cost per turn	Turns from $0.50
Short banter: 500 prompt, 150 reply	$0.000275	about 1,800
Persona chat: 3,000 prompt, 400 reply	$0.00115	about 430
Long scene with history: 8,000 prompt, 800 reply	$0.0028	about 175

Three habits keep the credit alive. Set max_tokens low while you experiment, because output costs four times as much per token as input. Trim chat history instead of resending every turn forever. And test with short prompts first, then move to your real persona once the plumbing works.

Plan the week as well. Spend the first day on connectivity and streaming, days two to four on prompts and personas, and keep the rest for load and error testing. When you finish, you will know your true cost per turn and can decide how much balance to add.

Making your first call

What is the quickest test?

Export your key as API_KEY and send one chat request. This one asks for a short in-character greeting and caps the reply at 80 tokens, so it costs a tiny fraction of a cent.

Copycurl https://api.uncensoredchatbotapi.com/v1/chat/completions \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"uncensored","max_tokens":80,"messages":[{"role":"user","content":"Introduce yourself as a cheerful ferry captain in two sentences."}]}'

Which base URL and model do I use?

The base is https://api.uncensoredchatbotapi.com/v1 and the only model id is uncensored.

Does it work with the OpenAI libraries?

Yes. The Python and Node packages accept a custom base URL, so point them at the address above and pass your key. See the docs for snippets.

I get 401. Is the trial broken?

A 401 means the key is missing or wrong. Check the header spells Authorization: Bearer followed by the key, and that you did not regenerate it, since regenerating invalidates the previous one immediately.

Keys and accounts

Can I have several keys?

Each account has a single key. You can regenerate it whenever you like, and the earlier key stops working at once.

Where do I keep the key safely?

On your backend, in an environment variable or secret store. Do not put it in web or mobile code. The web chat tutorial shows a small proxy that does this.

What if I lose the key?

Sign in and regenerate it. The new key is shown immediately and the old one stops working at once, so update your backend straight away.

Can teammates use my trial?

They can share your server-side proxy while you evaluate, but never hand out the key itself. Everyone behind it shares the same 300 requests per minute.

Questions about building a product

Is it okay to build an adult chatbot with it?

Yes, for audiences aged 18 and over. Lawful adult content, fiction and controversial topics are not refused. You should add your own age gate and rules, which the safety guide explains.

What is always off limits?

Sexual content involving minors is blocked every time with a 403, including in fiction and roleplay. Design your app so users understand that clearly.

Are chats used to train anything?

Prompts are not used for training.

Can the model draw pictures or speak aloud?

No. This is text only, with no images, audio, video, embeddings or fine-tuning, and one model.

Where can I read more on how the service is set up?

The hosted uncensored LLM overview covers what to expect, and the pricing page lists rates.

A seven-day evaluation plan

If you want a structured way to judge the service before topping up, follow this short plan. Each step uses only a few cents.

1. Day one: run the curl test above, then the same request through the Python or Node package with a custom base URL.
2. Day two: turn on streaming and check that text arrives progressively and that the last chunk reports usage.
3. Day three: drop in your real character prompt and run ten typical conversations. Note where the voice slips.
4. Day four: try tools if your product needs them, and confirm your code handles the tool_calls finish reason.
5. Day five: deliberately trigger errors: a wrong key, an oversized request, and a long loop to meet the rate limit. Confirm your UI copes.
6. Day six: calculate cost per turn from real usage and project a month of traffic.
7. Day seven: decide. If it fits, top up the prepaid balance before the credit expires so there is no gap.

Everything above works on a free account, so you lose nothing by testing thoroughly. A well-run week answers most questions that a sales call would.