NewPoint any agent at UberLLM — one key, every model, list price or less.
UberLLM

Janitor AI proxy setup with UberLLM (cheap uncensored models)

In a Janitor AI chat open API Settings, add a proxy configuration with Proxy URL https://api.uberllm.dev/v1/chat/completions, your UberLLM key and a model name such as deepseek/deepseek-v4.1-flash-uncensored, then Save Settings and refresh the page.

Get an API keySee live prices

Steps

  1. Make a key on your keys page and add credit.
  2. Open any chat on Janitor AI and click the gear icon at the top right, then API Settings.
  3. Choose the proxy option and click Add Configuration. (Not "janitor+ Router": that is Janitor's own paid service.)
  4. Fill in:
    • Config Name: UberLLM
    • Model Name: deepseek/deepseek-v4.1-flash-uncensored
    • Proxy URL: https://api.uberllm.dev/v1/chat/completions
    • API Key: your ull_… key, with no space before or after
    • Custom Prompt: leave it empty for now
  5. Click Add, then Save Settings.
  6. Refresh the whole page. Janitor only picks up a new proxy after a refresh.
  7. Send a message.

Unlike most apps, Janitor wants the full URL ending in /chat/completions. The short …/v1 URL does not work here.

Switching models

Edit the configuration and paste another id from the table below into Model Name, then Save Settings and refresh. The id must match exactly, including the part before the slash.

Keeping the bill small

  • Turn Text Streaming on (bottom of the ☰ menu) so long replies appear as they are written.
  • Long chats send the whole history every time. If a chat has hundreds of messages, lower the context in Janitor's generation settings or start a new chat with a summary.
  • Set a spend limit on the key (Edit on the keys page). Requests over it are refused and nothing is charged.

Troubleshooting

Janitor's own help center gives this order of fixes, and it solves most problems:

  1. Check the configuration says Active.
  2. Click Save Settings even if nothing changed.
  3. Refresh the page.
  4. Open the configuration again and send a test.

Then, by error:

  • "A network error occurred": the Proxy URL is wrong, or you did not refresh after saving. It must be exactly https://api.uberllm.dev/v1/chat/completions, with no extra slash at the end.
  • "Network Error" right after Check Key/Model: Janitor's help center says this button is unreliable. Save, refresh and send a real message instead.
  • 400: a typo or a space in the model name, URL or key.
  • 404: the model name does not exist. Copy it from the table below.
  • 402 / insufficient credits: add credit on the credits page.
  • Still unclear: open your browser's developer tools, Network tab, and look at the failed request; the reply text tells you what UberLLM said.
  • Works on one bot but not another: some bot creators block outside proxies for their character.

Keep reading

Models that work well on Janitor

Uncensored text models. Copy the id exactly into Model Name.

ModelModel id (paste this)Price now
DeepSeek V4.1 Flash Uncensoreddeepseek/deepseek-v4.1-flash-uncensored$0.1500 in · $0.6000 out / 1M
GLM-5.3 Flash Uncensoredz-ai/glm-5.3-flash-uncensored$0.1500 in · $0.5000 out / 1M
Venice Role Play Uncensoredvenice/uncensored-role-play$0.2405 in · $0.9620 out / 1M
Venice Uncensored 1.2venice/uncensored-1.2$0.0962 in · $0.4329 out / 1M
Qwen 3.6 Plus Uncensoredqwen/qwen3.6-plus-uncensored$0.4938 in · $2.96 out / 1M
Gemma 4 Uncensoredgoogle/gemma-4-uncensored$0.1313 in · $0.4040 out / 1M

Prices at the last hourly build, no fee added · every model →

If you want the strongest writing

Better memory for long stories; they may decline explicit scenes.

ModelModel id (paste this)Price now
DeepSeek V4.1 Flashdeepseek/deepseek-v4.1-flash$0.0070 in · $0.0783 out / 1M
Kimi K3moonshotai/kimi-k3$0.0032 in · $0.2475 out / 1M
GLM 5.3z-ai/glm-5.3$0.0023 in · $0.0469 out / 1M
Gemini 3.5 Flashgoogle/gemini-3.5-flash$0.0592 in · $0.3555 out / 1M
Claude Sonnet 5.5anthropic/claude-sonnet-5.5$0.7610 in · $3.80 out / 1M

Prices at the last hourly build, no fee added · every model →

FAQ

What is the UberLLM proxy URL for Janitor AI?

https://api.uberllm.dev/v1/chat/completions — the full URL. Paste your UberLLM key as the API key and an UberLLM model id as the Model Name, then Save Settings and refresh.

Which models can I use on Janitor AI through UberLLM?

Any UberLLM chat model. Uncensored text models such as deepseek/deepseek-v4.1-flash-uncensored and z-ai/glm-5.3-flash-uncensored are the usual pick for roleplay; the guide lists the ones online now with prices.

Why does Janitor say network error after I added the proxy?

Usually the URL is not exactly https://api.uberllm.dev/v1/chat/completions, or the page was not refreshed after Save Settings. The Check Key/Model button can also show a false network error.