JanitorAI Proxy (Openrouter + Chutes) Setup [Updated 6-14]

[YOU CAN REUSE/REPOST THIS AS LONG AS IT LINKS BACK TO THIS PASTEBIN OR GIVES CREDIT]


[Alright, chutes has gone down the shitter in terms of quality. While my time using it was brief, I found it didn't handle my bots/sims nearly as well as non-chutes version. I wouldn't recommend it, but it's there if you wish to use it]


-----------------

A new service was brought to my attention, which is chutes.ai. It does not have a limit, unlike Openrouter.ai, at the cost of being slightly more complex to set up and worse performance.
Here's a post that goes over it pretty well: https://www.reddit.com/r/JanitorAI_Official/comments/1ju5vih/visual_guide_for_deepseek_users_via_chutesai_full/?rdt=44947


While the directions there are pretty accurate, some parts of it isn't. So, I will be reposting the guide on that post here with slight revisions to make it easier for others to understand. The following is 80% NOT made by me.


Setting up an Chutes Account for AI chat: Deepseek V3 JanitorAI Version

Go to https://chutes.ai/app and click "Create an account".

Enter any name you want.

Click "Create my Account".

Fingerprint: Don’t type anything...just COPY the one provided. [The fingerprint system is simply a big blurg of text that acts as your password] SAVE THIS IN A NOTE! If you lose it, you can’t log back in.

After you log in, click the API selection in the top middle of the screen next to My Chutes and Stats. This is where you'll generate you're key

Type anything in the box (e.g., "DeepSeekAPI") → Click "+ Create".

Just like Openrouter, you'll have to save this API key. I recommend saving it separately from the fingerprint or labeling them separately.

When you have the key, simply go into Janitor AI and input these 3 things into the Proxy. [Don't know how to do that? Look at steps 10-14 in the Openrouter Guide below. Don't want deepseek V3 and instead another model? Use the post linked above to see how to get model names for another LLM, but I'd highly recommend deepseek V3.]

API Key: (Your key)
Model: deepseek-ai/DeepSeek-V3-0324
Proxy URL: https://llm.chutes.ai/v1/chat/completions

After this, do NOT select check API, you will likely get an error. Go down and click save settings, then reload the page, THEN select check API to make sure it's properly connected.

NOTE: If something is wrong or it's not working, check the bottom of the page on the ISSUES tab. If those don't work, make sure to check the reddit post, as they most likely have the solution for you there.


--------------------


Setting up an Open router account for AI chat: Deepseek V3 JanitorAI Version

1. Go to https://openrouter.ai/

2. Sign in to your email in the top right corner

3. Hover over profile picture (top right) and select the "keys" dropdown

4. Select the "create key" button

5. Name it to whatever you want, for this example I will call it Deepseek V3 to keep track of my keys and to remember which model goes with which key
- Credit limit is optional and, if you plan to not use paid models and use a free one like deepseek V3, you can leave it alone

6. SAVE THE ENTIRE KEY TO SOMEWHERE SAVE AND EASILY RETRIEVEABLE! (For example, a notepad on your desktop) You will NOT be able to copy your key again from the site.

7. Before you leave the settings/key page, go to the menu on the left side and select the "privacy" tab. Make sure that, under "Free Models", that the "Enable training and logging (chatroom and API)" Option is turned ON. You need this on for Janitor AI to get endpoints and send data back and forth. You will not be able to chat with a proxy otherwise

8. Go to models and find a model for you. For this example, we will be using my favorite and recommended model, DeepSeek V3 0324 (FREE), which can be found here: https://openrouter.ai/deepseek/deepseek-chat-v3-0324:free

9. That link is what you'll want to copy, but you can also click the clipboard button under the main name. This name of the model is important, keep it.

--

6-14: After plenty of testing, I found that I preferred Deepseek R1 0528 over Deepseek v3, as it seems to give more in-depth answers (at the cost of being a bit much/flowery sometimes) and not sticking in reasoning randomly like Deepseek v3. Simply follow the steps above, but put these in instead:

Model: deepseek/deepseek-r1-0528:free

Proxy URL: https://openrouter.ai/api/v1/chat/completions

Key: Your personal key (make sure to make this its own key to avoid issues!

You should always try and have one free model per key, to avoid any potential issues.

--

###


OPTIONAL STEP: SETTING UP CHUTES.AI AS A FALLBACK
If you wish to use mainly use openrouter, and don't want to switch API information every time it runs out, you can use chutes.ai as a fallback integration instead, so you'll never have to switch every time you reach your limit on openrouter

1. Set up a full Chutes account. This can be done at the top.
2. Sign into Openrouter. Hover over your profile in the top right, then go to settings
3. Click on Integrations
4. Select Chutes on the list
5. Put in the key for Chutes, along with turning both the 'Enabled' selection and the 'Use this key as a fallback' selection
6. Save it, and you should be good.

Other AI Sites can be used as a fallback to Openrouter, but I'll only be going over Chutes. With this system done, you WILL NOT need chutes.ai Model and Proxy URL. Instead, simply use the same key, model, and Proxy URL as Openrouter and it will automatically start using Chutes when Openrouter has run dry.


OPTIONAL STEP: Disabling reasoning on openrouter
Sometimes the model reasons instead of actually providing a response. Doing this can fix this.
[NOTE: This is from someone else, and I haven't tested it, so if this screws things over, just undo the steps]

1. Go into settings
2. In the middle, there will be a selection for 'Ignored Providers'
3. Search and put in Azure and Targon


OPTIONAL STEP: Increasing daily limit through money

Many people, myself included, don't wish to pay. However, if you have 10 credits stored in your account, which I BELIEVE to be $10 USD but I am uncertain, the daily limit goes from around 50 to 1000 requests per day for all free models. This is something I won't do, but if you really want an option for daily limits, you can. MAKE SURE to set your key to not use any credits so you don't use up your paid credits.


###


SETTING UP IN JANITORAI

10. Go into JanitorAI and go to a proxy enabled character

11. Click the 3 bars on the top left, and go to API settings, then Proxy

12. From top to bottom, the first text field you want to put in the name of the model you're going to be using. For deepseek v3, it is the link from above. (deepseek/deepseek-chat-v3-0324:free)

13. The 2nd field is for Openrouter itself. No matter what model you use, if you get it from openrouter, it will use this: https://openrouter.ai/api/v1/chat/completions

14. Finally, the 3rd field is for the API key you saved locally or in a secure location. After putting it in, all you have to do is save [Press the Save settings button], reload, and it should be applied! (Make sure to do this BEFORE selecting the check API button) MAKE SURE TO CHANGE GENERATION SETTINGS! They get reset after applying a proxy.

While V3 is great, do keep in mind that there is a limit on generations from the model, but it's pretty generous. Refer to the Setting up Chutes.ai and the fallback if you wish to use openrouter and don't want to suffer from limits.


POSSIBLE ISSUES
If you are getting a networking error, or it simply is not working, some possible solutions can be:
Copy and pasting the model name and Proxy URL, since typing it can cause issues
Closing and Reopening JanitorAI tabs
Make sure you Save Settings and reload the page BEFORE checking the API.


Personal notes as of 6-15:
The model I will personally prefer is Deepseek R1 0528. It has much more grittier and overall use for roleplay, and handles prompting very well. Best part it, it doesn't have a immense positivity bias unlike most models, so it's great for that.

To get your best worth out of it, I would recommend using Deepseek as a starter and when JLLM can't get the scenario right or doesn't have correct knowledge. Otherwise, use JLLM for general use and conversation. Despite it being slow and not as complex, it does roleplay the best, and has the best dialogue (from my experience)


Thank you for your time.

Made by NullSeu