EN ▾

Open WebUI OpenAI Compatible API Setup in 5 Minutes

To use open webui openai compatible api settings, you add one connection with a base URL and a key, and the model appears in the picker. Here is the exact setup for the Abliterated API, from Docker to first chat.

Updated

Chat interface linked by a glowing cable to a hosted API server, representing an Open WebUI connection

Key points

  • Base URL is https://api.abliteratedapi.com/v1 and the model ID is uncensored.
  • Add it in the admin connections settings, or pass OPENAI_API_BASE_URL and OPENAI_API_KEY to Docker.
  • Test the key with curl first, so you can tell a key problem from a UI problem.
  • Open WebUI is only the front end: you pay per token for what the model reads and writes, and no card is needed to start.

The short version

If you want the open webui openai compatible api setup in one paragraph: open the admin settings, add an OpenAI API connection, set the URL to https://api.abliteratedapi.com/v1, paste your key, save, and pick the model called uncensored. That is all. Open WebUI talks to anything that implements the OpenAI chat completions format, and this API does.

The service is a hosted, OpenAI-compatible text API running one open-weight model, tuned to answer without content refusals for lawful adult use. It is run by an independent provider and is not the official API of any model lab. Open WebUI is just the chat front end you point at it.

Before you begin

  • An API key. Create one on the API key page. Sign-up is with Google or with email and password. New users can claim $0.50 of trial credit with one click on that page, valid for 7 days and no card required.
  • A running Open WebUI. A Docker install or a pip install both work. The steps below cover Docker and the settings screen.
  • Network access from the machine running Open WebUI to api.abliteratedapi.com over HTTPS.

Check the key before touching the UI. If this call returns a list containing uncensored, the key and URL are good:

curl https://api.abliteratedapi.com/v1/models   -H "Authorization: Bearer $ABLITERATED_API_KEY"

Then confirm a real completion works:

curl https://api.abliteratedapi.com/v1/chat/completions   -H "Authorization: Bearer $ABLITERATED_API_KEY"   -H "Content-Type: application/json"   -d '{"model":"uncensored","messages":[{"role":"user","content":"Say hello in five words."}]}'

Method 1: add the connection in the settings

This is the easiest route and works on an existing install. Menu labels move around between Open WebUI versions, so the names below may differ slightly.

  1. Sign in to Open WebUI as an admin and open the Admin Panel, then Settings, then Connections.
  2. Find the OpenAI API section and add a new connection with the plus button.
  3. In the URL field enter https://api.abliteratedapi.com/v1. Stop at /v1. Do not add /chat/completions; Open WebUI appends the path itself.
  4. Paste your key into the key field. It is sent as Authorization: Bearer <key>.
  5. Use the verify button next to the fields if your version has one, then save.

Open a new chat and open the model selector. The model uncensored should be in the list, because Open WebUI reads it from GET /v1/models. If the list stays empty, go to the troubleshooting section below.

Method 2: set it at startup with Docker

For a fresh install, pass the endpoint as environment variables. Open WebUI reads OPENAI_API_BASE_URL and OPENAI_API_KEY when it starts.

# base URL: https://api.abliteratedapi.com/v1   model to pick in the UI: uncensored
docker run -d -p 3000:8080   -e OPENAI_API_BASE_URL="https://api.abliteratedapi.com/v1"   -e OPENAI_API_KEY="$ABLITERATED_API_KEY"   -v open-webui:/app/backend/data   --name open-webui --restart always   ghcr.io/open-webui/open-webui:main

Then visit http://localhost:3000, create the first account (it becomes the admin) and start a chat. The named volume keeps your chats and settings when you update the container.

One thing catches people out: many Open WebUI settings are saved to its database after the first launch, and the saved value can win over a later environment variable. If you change the key or URL in Docker and nothing happens, change it in the Connections screen instead, or reset the stored setting.

Settings worth adjusting

Once chatting works, a few options make the experience better and keep the bill predictable.

Use the context window sensibly

The context window is 100,000 tokens, shared between prompt and reply. Open WebUI resends the chat history on every message, so a long conversation gets more expensive with each turn. At $0.15 per 1M input tokens, a 20,000-token history costs $0.003 to send, which is cheap, but it adds up across thousands of messages. Start a new chat for a new topic.

Streaming and sampling

Open WebUI streams replies by default, and the API supports streaming over SSE. In the advanced parameters you can set temperature, top_p, stop sequences, a seed for repeatable output and the penalties. These are passed through to the model. Leave max_tokens empty unless you want a hard cap, since it is optional.

Background tasks

Open WebUI can make extra model calls in the background for things like naming a chat. Each one is a real request on your balance and counts toward the limits. If you notice calls you did not type, look at the task model settings and point them at a model you are happy to pay for, or switch the helper features off.

Presets and system prompts

Open WebUI lets you save a model preset in its workspace section: a name, a base model, a system prompt and default parameters. Create one on top of uncensored for each job you do often, such as an editing assistant, a code reviewer or a character for an interactive fiction session. A short system prompt that states the role, the tone and the format of the replies does more than long instructions pasted into every chat.

This is where the model's profile helps. It is an open-weight model tuned to answer without content refusals for lawful adult use, which suits fiction, roleplay and companion-style apps for adults, where a refusal in the middle of a scene breaks the flow. It does not change the hard limit: no sexual content involving minors, ever. Keep presets and shared instances within that rule and within the law where you live.

If other people use your Open WebUI instance, remember that they all share your key and its limits. A full key allows 8 concurrent requests, so a handful of active users is fine, while a large group chatting at once can hit the cap. Giving each person or team their own key spreads the load and makes the balance easier to track.

What it costs to run

You pay for tokens only, from prepaid credit: $0.15 per 1M input tokens and $0.50 per 1M output tokens. Failed or refused requests are not charged and the credit never expires. A typical chat message with 3,000 tokens of history and a 400-token reply costs $0.00045 + $0.0002 = $0.00065, so $10 of usage is about 15,000 messages.

Top-ups run from $20 to $500 in any whole amount, with +5% from $50, +10% from $100 and +15% from $200. You can pay by crypto (USDT on TRC20, USDC on Base). See the pricing page for the details. The remaining balance is returned in the X-Balance-USD header of each API response, if you want to monitor it from a script.

Troubleshooting

SymptomLikely causeFix
Model list is emptyWrong URL or keyRun the curl test for /v1/models. URL must end in /v1.
Connection refused or timeout from DockerNo outbound HTTPS from the containerCheck DNS, proxy and firewall settings on the host.
Auth error on every messageKey pasted with spaces, or old keyRe-copy the key from the key page and save again.
Errors when several people chat at onceRate limitsA full key allows 8 concurrent requests and 300 per minute; a trial key allows 2 and 60.
Long chat stops workingContext window fullThe window is 100,000 tokens. Start a new chat or summarize.

If the curl test passes but Open WebUI fails, the issue is in the front end settings. If curl fails, check your key and balance first. The docs quickstart lists every request parameter, and the article on what the uncensored chat API is and is not explains the model's scope, including the hard limit: no sexual content involving minors.

Questions and answers

What URL do I enter in Open WebUI?

Use https://api.abliteratedapi.com/v1 and nothing after it. Open WebUI adds /chat/completions and /models on its own.

Which model name do I choose?

The model ID is uncensored. It appears in the model picker once the connection is saved and the models list loads.

Do I need a card to try it?

No. New users can claim $0.50 of trial credit with one click on the API key page after sign-up. It is valid for 7 days and needs no card.

Can I run it alongside local models?

Yes. Open WebUI supports several connections at once, so this endpoint can sit next to local models and each shows up in the same model picker.

Are my chats used for training?

No. Prompts sent to the API are not used for training. Chat history stored by Open WebUI stays in your own installation.

Your key is one form away

Create an account, copy the key, change the base URL. That is the whole setup.

Get API key