LibreChat with an uncensored model

Where the settings are

LibreChat reads librechat.yaml from the project root, and a custom endpoint is one item under endpoints.custom: a name, a key, a base URL and a model list. The key goes in .env and the YAML references it; LibreChat’s docs call the environment-variable reference the recommended form for apiKey. Docker users mount the file through docker-compose.override.yml and restart.

Steps

1. Create a key in the console and put it in .env next to docker-compose.yml:

2. Add the endpoint to librechat.yaml. The ids under models.default are a starting list; any id on the models page goes in the same way:

3. Keep the mount in the override file, then restart:

docker compose down && docker compose up -d reloads the file. The endpoint then appears in the selector under the name you gave it, with the models in models.default.

What each setting does

name

UncensoredLLM

the title in the endpoint selector; required and unique

apiKey

${UNCENSOREDLLM_API_KEY}

reads the sk-ullm- key from .env; user_provided makes each user paste their own

baseURL

https://api.uncensoredllm.ai/v1

with the /v1; LibreChat appends /chat/completions itself, so directEndpoint stays unset

models.default

the ids above

the picker, as long as fetch is false; required

models.fetch

false

true would load every model on the key into the picker

titleConvo, titleModel

glm-5.3-flash-uncensored

the automatic chat title and the small model that writes it

modelDisplayLabel

UncensoredLLM

the label beside the AI icon in the chat; "AI" by default

Picking a model

LibreChat sends the whole conversation on every turn in the OpenAI format, and every build on the models page answers it. The cost column is one turn of 2,000 tokens in and 500 out at the catalog prices on 2026-11-09.

A first reply on GLM-5.3 Uncensored is $0.00099; with titleConvo on, each new chat adds one title request of about 1,000 tokens in and 20 out, $0.0003 on GLM-5.3 Uncensored and $0.00011 on GLM-5.3 Flash Uncensored, which is why the example sets titleModel.

Check it

From the machine that runs LibreChat, list the models the key can call; every id under models.default should be in the answer:

Then send one streamed request the way LibreChat does. -i prints the headers: x-uplink-request-id names the request for support.

If both work and LibreChat does not, the problem is in the YAML or the mount. Check the container log after docker compose up.

Errors and fixes

• The endpoint is missing from the selector: the file is not mounted, or the YAML did not parse. The container log after docker compose up names the line.
• 401 invalid_api_key: the key is wrong or revoked, or the variable name in .env does not match the one in apiKey.
• 402 insufficient_balance: the balance does not cover the request. Top up in the console. 402 monthly_limit_reached means the key hit the monthly spend limit you set on it.
• 404 model_not_found: an id in models.default is not on the models page, or a display name was typed where the id belongs.
• 429 rate_limited: the key hit its rate limit; every user of one endpoint shares one key and one limit. The response carries retry-after.
• Every model on the key shows up in the picker: fetch is on. Set it to false and keep the list in models.default.

Sources

Field names, the env-var reference, the baseURL convention and the mount are from LibreChat’s own docs, read 2026-11-06: Custom Endpoint Object Structure and Custom Endpoints quick start. Prices are from the live catalog at render.

Questions