This tutorial covers a brief summary on LLMs, proxy LLMs, the warnings/side effects, tutorial on how to set up proxy models, recommended models (free and paid), Meganova v.s. Openrouter proxy models, API settings, prompts, suggestions, etc.

Location Map
Brief summary, warnings, and suggestions - Just below the map!
Tutorial - Just below the summary!
API Settings - Scenario section
Prompts - Personality section
Recommended Models - Initial Messages
Initial messages - 1: Meganova models
                             2: Openrouter models

Summary
- Alternative LLMs can be easily accessed in place of the JLLM (default LLM) and are usually deployed by users who are interested in engaging in a more immersive roleplay experience.

- Beware of proxies! The ones I listed (Meganova and Openrouter) are generally considered safe and have been tested by a majority of Janitor AI creators. However, there are unfortunately many “scraper” proxies online that may leak character definitions or even leak IPs. Your safety is a priority, and I highly recommend double-checking what you use.

- Please note that neither I nor the AI models can read your mind. I highly recommend testing a lot of the free/cheap models (like the one's provided in the initial messages) to conclude which model best fits your preferences. Use chat summaries if you're struggling with bot memory, mess around with the generation settings if the bot isn't behaving how you'd like, use API prompts that are specifically tailored to fix common annoying AI cliches (e.g. 'her breath hitched, just a fraction.') Keep testing until your satisfied with your RP experience.

Tutorial for Proxies
1. Click on any bot that allows proxies (like this one)
2. Head to the top right corner and click this button

3. Next to the Janitor AI LLM, there should be a tab that says ‘Proxy’ – Click it

4. Scroll down to ‘Proxy Configurations’ and click it. Then, click ‘+New’

5. Go to https://www.meganova.ai/ or https://openrouter.ai/ and get an API key (they’re free!)
This key can be used for multiple LLMs as long as they are under the same proxy. Do not share this key with anyone.

6. Find a desired LLM offered in your proxy of choice (recommendations in initial messages), and put the exact model name in the model section

Example of what an exact model name looks like


7. Enter your proxy URL (https://inference.meganova.ai/v1/chat/completions or https://openrouter.ai/api/v1/chat/completions) and the API key you retrieved earlier in under the model name section

8. Hit save and you’re finished!

This tutorial was tested on PC. I have no idea how this looks on mobile but I'm sure it works the same