# 🌸**𝐒𝐚𝐤𝐮𝐫𝐚 𝐏𝐫𝐨𝐱𝐲 𝐆𝐮𝐢𝐝𝐞**🌸
**The guide is written for Janitor, but all proxies can be used for sites where proxies are supported (Chub, Sauspaun,Silly Tavern and etc)**
## ⛩️ [ ***Material Authors*** ] ⛩️
🌸 *Before the guide begins, let's say thank you to all the people who helped the community find proxies. The best way to thank the authors is to support them!* 🌸
*These people have nothing to do with this guide. I just took materials from their channels. If something is wrong, blame [Ejenenek](https://janitorai.com/profiles/8ed821cf-7c79-4496-9394-ae956cd72fbe_profile-of-ejenenek)* <-
---
🎋 **Special Thanks & Sources**
* ✨ **A huge thank you for the materials to:** [**iigrrd** (Telegram)](https://t.me/iigrrd) // [**Janitor Profile**](https://janitorai.com/profiles/e80bc2de-8e13-406b-9de7-3b09076cb294_profile-of-iigrrd)
* 💝 **Warmest thanks for your help:** **Dillen**
[[ Janitor ]](https://janitorai.com/profiles/3978d4e3-bfd6-419a-a88a-b6336357cf5e_profile-of-dillen) // [[ Saucepan ]](https://saucepan.ai/u/Dillen)
**📚 Other Materials used:**
* 💮 [**Unknown_Nanachi**] [Janitor Ai]( https://janitorai.com/profiles/521244d9-de98-4215-bc65-a8d911dda1ce_profile-of-unknown-nanachi)
* 💮 [AI and Drama](https://t.me/AIandDrama) | [Trigger Tourist](https://t.me/Trigger_Tourist)
* 💮 **Eva & Allora:** [TG](https://t.me/evaandallora) | [Janitor](https://janitorai.com/profiles/a580a42e-ab10-4ee6-8ca7-2191568f2d3a_profile-of-eva-porsche)
* 💮 **Honeydew:** [TG](https://t.me/honeydew_ew) | [Janitor](https://janitorai.com/profiles/b2ec0321-8b8d-48e9-87e5-32b2d4bbdf73_profile-of-honeyd-ew)
* 💮 **finhead94** [Janitor Ai](https://janitorai.com/profiles/c83f4b3b-8823-40e2-ab32-3383e167f24f_profile-of-finhead-94)
* 💮 **Dartjuk** [Janitor Ai](https://janitorai.com/profiles/0e5a58ab-e0f7-40c3-b4a3-0eb10f29aac7_profile-of-dartjuk)
---
# 🏮 **IMPORTANT NOTICE** 🏮
*Think twice before using any proxy (whether it's paid or free).*
***DO NOT rush to buy a proxy, ESPECIALLY if it just recently appeared.***
> **There are a number of dangers when using a proxy:**
> 1. 🎏 **Your logs** (chats) can be and **likely are being read**.
> 2. 🀄 When purchasing, they may **steal your data** (card details).
> 3. 🍙 Proxy may use **stolen keys**, leading to an account ban.
> 4. 🏯 **Your data** (IP, location, chats) may be leaked for shady purposes.
---
# 🎋 **What to do?** 🎋
1. **Read the terms of service** (if they don't exist, reconsider using that proxy).
2. **IF YOU ARE BUYING SOMETHING, THINK 20 TIMES.**
3. **DO NOT rush** to buy a subscription from "one-day" proxies.
4. If you want to buy, look for long-standing providers or go official (**DeepSeek**).
---
---
# 🏯 What do you need to know before setting up a proxy?🏯
---
**🈷️General collection of guides** [(How to make a bot right,what is memory, promts, extensions, and a bunch of other stuff)](https://janitorai.com/characters/23d18d3d-cf16-4c6a-a7e4-47bfcc543cec_character-%D1%81%D0%B1%D0%BE%D1%80%D0%BD%D0%B8%D0%BA-%D1%81%D0%BE%D0%B2%D0%B5%D1%82%D0%BE%D0%B2-%D1%81%D0%B1%D0%BE%D1%80%D0%BD%D0%B8%D0%BA-%D1%81%D1%82%D0%B0%D1%82%D0%B5%D0%B9-%D0%B4%D0%BB%D1%8F-%D0%BD%D0%BE%D0%B2%D0%B8%D1%87%D0%BA%D0%BE%D0%B2-%D0%B8-%D1%81%D1%82%D0%B0%D1%80%D0%B8%D1%87%D0%BA%D0%BE%D0%B2-%D1%80%D1%83%D0%BA%D0%BE%D0%B2%D0%BE%D0%B4%D1%81%D1%82%D0%B2%D0%BE-%D0%BA%D0%B0%D0%BA-%D1%81%D0%BE%D0%B7%D0%B4%D0%B0%D1%82%D1%8C-%D0%B1%D0%BE%D1%82%D0%B0-%D0%BF%D0%BE%D0%BC%D0%BE%D1%89%D1%8C-%D0%BD%D0%BE%D0%B2%D0%B8%D1%87%D0%BA%D0%B0%D0%BC)
---
# 🍙Basic terms
### 1. PROXY
A "bridge" or intermediary server that connects JanitorAI to external AI services. It allows you to use models that are not natively hosted on the site by routing your messages through a custom URL.
### 2. PROVIDER
The service or company that hosts the AI models and gives you access to them (usually via an API key).
* Examples: OpenRouter, Chutes, or the official DeepSeek API.
### 3. MODEL
The actual "brain" or software program that generates the text. Different models have different styles, logic, and creativity levels.
* Examples: DeepSeek-V3 (smooth/creative), DeepSeek-R1 (logical/reasoning), or JLLM (default).
---
## 🀄Where can I find a proxy? Where do I find one?
> 1. Go to any bot chat
>2. Click on the purple box in the corner **using proxy** or **using Janitor**
>3. Chose **Proxy**
>4. Creaty **New Configuration**
>5. Fill in in accordance with the instructions
---
## 🍶 Optimized RP Settings
For a better roleplay experience, you can slightly tweak these settings:
### 1. Basic Settings
* **Temperature:** **0.6 – 0.8**. If you set it to 1.0 or higher, the bot tends to "go crazy" and lose coherence.
* **Tokens:** **0**. Setting this to 0 (unlimited/max) ensures the bot finishes its sentences instead of cutting off mid-reply.
* **Context Size:** Leave as is (default). Usually around **40k** memory tokens.
---
### 2. Advanced Controls

* **Banned Words (Banned Tokens):** Add repetitive or annoying words here (e.g., "my," "grabbed your wrist," etc.). Once added, the bot will stop using these specific phrases in the RP.
* **[ Prefill]:** This is "forced" information. It is the first thing the bot reads and what it will prioritize 100%. Use this to mandate specific behaviors (e.g., forcing NSFW content or adding info-boards if the bot was previously refusing).
* **[ Prompt ]:** The primary instructions for the bot on *how* to write (style, tone, and formatting rules).
**For Proxy**
**For LLM**
---
**Where to look for prompts?**
[**My Collection**](https://janitorai.com/characters/a6d9aa74-32d0-4929-98ac-d2572107af28_character-a-collection-of-various-promts-for-any-models)| [**Janitor AI Discord**](https://discord.gg/janitorai)
---
### 🌸PROXY CONNECTION GUIDE!
> *A complete guide to connecting Free Proxies for your roleplay experience.*
---
### 🗻 OpenRouter
Register: [OpenRouter.ai](https://openrouter.ai/)
Copy your API KEY from the **Keys** section.
Choose your model: Look for the `:free` tag (50 messages/day).
Setup in JanitorAI: **Bot** -> **Proxy** (top right corner) -> **Create New Configuration**.
> **Configuration Name:** `Any name`
> **Model:** `Choose from the list below`
> **URL:** `https://openrouter.ai/api/v1/chat/completions`
> **API KEY:** `[Your copied key]`
---
### 🌸 Several Models
Paste these into the **Model** field (URL stays the same):
* 🌸 `tngtech/deepseek-r1t2-chimera:free`
* 🌸 `moonshotai/kimi-k2:free`
* 🌸 `deepseek/deepseek-r1-0528:free`
---
### 🌊 LLM7
1. **Register:** [token.llm7.io](https://token.llm7.io/)
2. **Copy your API KEY.**
3. **Choose your model:** [llm7.io](https://llm7.io/)
4. **Setup in JanitorAI:** Bot - Proxy (top right corner) - Create New Configuration.
* **Configuration Name:** Any name.
* **Model:** Your
* **URL:** `https://api.llm7.io/v1/chat/completions`
* **API KEY:** Your copied key.
#### 🌸 Several Models:
*Paste these into the Model field (URL stays the same):*
(Just a few models, see the rest on the website. They are constantly changing.)
• `deepseek-v3-0324`
• `deepseek-r1-0528`
• `mistral-small-3.1-24b-instruct-2503`
• `gpt-4.1-nano-2025-04-14`

[🎥 Video Guide](https://www.tiktok.com/t/ZT6jXqjky/)
---
### 🌌 NEBULA
1. **Register:** [nebulablock.com](https://www.nebulablock.com/)
2. **Copy API Key:** [From here](https://www.nebulablock.com/apiKeys)
3. **Setup in JanitorAI:** Bot - Proxy - Create New Configuration.
* **Model:** `deepseek-ai/DeepSeek-R1-0528-Free`
* **URL:** `https://inference.nebulablock.com/v1/chat/completions`
* **API KEY:** Your copied key.
> ⚠️ **IMPORTANT:** To use DeepSeek on Nebula, you must **link a card**. No charges will be made. Other models work **without a card**. This is due to high demand on DeepSeek.
#### 🌸 Several Models:
*Paste these into the Model field (URL stays the same):*
(Just a few models, see the rest on the website. They are constantly changing.)
* ` deepseek-ai/DeepSeek-R1`
* ` deepseek-ai/DeepSeek-V3-0324`
* `deepseek-ai/DeepSeek-R1-0528`
* ` deepseek-ai/DeepSeek-R1-Free`
* ` deepseek-ai/DeepSeek-V3-0324-Free`

[🎥 Video Guide](https://www.tiktok.com/t/ZT6jXRSh7/)
---
### 🌿 SILICONE FLOW
1. **Register:** [siliconflow.com](https://www.siliconflow.com/)
2. **Copy API Key.**
3. **Setup in JanitorAI:** Bot -Proxy - Create New Configuration.
* **Model:** `deepseek-ai/DeepSeek-R1`
* **URL:** `https://api.ap.siliconflow.com/v1/chat/completions`
* **API KEY:** Your copied key.
#### 🌸 Several Models:
(Just a few models, see the rest on the website. They are constantly changing.)
*Paste these into the Model field (URL stays the same):*
* `deepseek-ai/DeepSeek-R1`
* `deepseek-ai/DeepSeek-V3`
* `moonshotai/Kimi-K2-Instruct`
* `Qwen/Qwen3-32B`
* `MiniMaxAI/MiniMax-M1-80k`
[🎥 Video Guide](https://www.tiktok.com/t/ZT6j9RNM5/)
---
### 🏮 GMICLOUD
1. **Register:** [gmicloud.ai](https://www.gmicloud.ai/)
2. **Create & Copy API Key:** [Go here](https://console.gmicloud.ai/user-setting/organization/api-key-management)
3. **Choose Model:** You can find models in the [LLM tab](https://console.gmicloud.ai/?group=LLM).
4. **Setup in JanitorAI:** Bot Chat - Purple Proxy Button - Proxy - Create New Configuration - Save.
5. **Fill as follows:**
* **Model:** Your chosen model (e.g., `deepseek-ai/DeepSeek-R1-0528`)
* **URL:** `https://api.gmi-serving.com/v1/chat/completions`
* **API KEY:** Your copied key.
#### 🍡 Free Models:
*(Just a few models, see the rest on the website. They are constantly changing.)*
*Paste these into the Model field (URL stays the same):*
* `deepseek-ai/DeepSeek-R1-0528`
* `meta-llama/Llama-3.3-70B-Instruct`
* `Qwen/Qwen3-32B-FP8`
* `moonshotai Kimi-K2-Instruct`
[🎥 Video Guide](https://www.tiktok.com/)
---
### 🌿 NOVITA
1. **Register:** [novita.ai](https://novita.ai)
2. **Get API Key:** [Go here](https://novita.ai/settings/key-management)
3. **Choose Model:** Scroll to the bottom and copy from [here](https://novita.ai/docs/guides/llm-reasoning).
4. **Setup in JanitorAI:** Bot Chat - Purple Proxy Button - Proxy - Create New Configuration - Save.
5. **Fill details:**
* **Configuration Name:** Any name you like.
* **Model:** Your chosen model (e.g., `deepseek/deepseek-r1-distill-qwen-32b`)
* **URL:** `https://api.novita.ai/v3/openai/chat/completions`
* **API Key:** Paste your copied key.
#### 🌸 Free Models:
*(Just a few models, see the rest on the website. They are constantly changing.)*
*Paste these into the Model field (URL stays the same):*
* `deepseek/deepseek-r1-turbo`
* `deepseek/deepseek-r1-distill-qwen-32b`
* `qwen/qwen3-235b-a22b-fp8`
* `meta-llama/llama-4-maverick-17b-128e-instruct-fp8`
---
### 🏮 AIRFORCE
1. **Register:** [panel.api.airforce/signup](https://panel.api.airforce/signup/)
2. **Create & Copy API Key.**
3. **Choose Model:** Click on the `=` icon to browse.
4. **Setup in JanitorAI:** Bot Chat - Purple Proxy Button - Proxy - Create New Configuration - Save.
5. **Fill as follows:**
* **Model:** Your chosen model name.
* **URL:** `https://api.airforce/v1/chat/completions`
* **API KEY:** Your copied key.
#### 🍡 Free Models:
*(Just a few models, see the rest on the website. They are constantly changing.)*
*Paste these into the Model field (URL stays the same):*
* `deepseek-terminus`
* `deepseek-v3.2-thinking`
* `claude-sonnet-4.5`
* `llama-4-scout`
---
### 🎐 ROUT.MY
1. **Register:** [rout.my](http://rout.my)
2. **Discord Setup:** Click **Dashboard** and join their Discord.
3. **Get Key:** In Discord, go to the `#My key` channel to generate your API key (requires email).
4. **Choose Model:** Go to `#models`, click the button, and select **Free**. Copy the model name.
5. **Setup in JanitorAI:** Bot Chat - Purple Proxy Button - Proxy - Create New Configuration - Save.
* **Model:** Your chosen model.
* **URL:** `https://api.rout.my/v1/chat/completions`
* **API KEY:** Your copied key.
#### 📜 Free Models:
*(Just a few models, see the rest on the website. They are constantly changing.)*
*Paste these into the Model field (URL stays the same):*
* `anthropic/claude-haiku-4.5`
* `openai/gpt-5`
* `google/gemini-2.5-pro`
* `deepseek/deepseek-v3.2-exp`
[🎥 Video Guide](https://www.tiktok.com/)
---
### 🌊 NAGA.AC
1. **Register:** [naga.ac](http://naga.ac)
2. **Copy API Key:** [Go here](https://naga.ac/dashboard/api-keys)
3. **Choose Model:** Look for models with the **Free** tag [here](https://naga.ac/models).
4. **Setup in JanitorAI:** Bot Chat - Purple Proxy Button - Proxy - Create New Configuration - Save.
* **Model:** Your chosen model.
* **URL:** `https://api.naga.ac/v1/chat/completions`
* **API KEY:** Your copied key.
#### 🐚 Free Models:
*(Just a few models, see the rest on the website. They are constantly changing.)*
*Paste these into the Model field (URL stays the same):*
* `llama-4-scout-17b-16e-instruct:free`
* `gpt-4o-2024-11-20:free`
* `claude-3.5-haiku-20241022:free`
* `deepseek-v3.2-exp:free`
---
### 🌌 LITEROUTER
1. **Register:** [literouter.com](https://literouter.com/)
2. **Copy API Key:** [Go here](https://literouter.com/api_dashboard.php)
3. **Choose Model:** Look for the **Free** tag [here](https://literouter.com/api_dashboard_models.php).
4. **Setup in JanitorAI:** Bot Chat - Purple Proxy Button - Proxy - Create New Configuration - Save.
* **Model:** Your chosen model.
* **URL:** `https://api.literouter.com/v1/chat/completions`
* **API KEY:** Your copied key.
#### 🎋 Free Models:
*(Just a few models, see the rest on the website. They are constantly changing.)*
*Paste these into the Model field (URL stays the same):*
* `deepseek-free`
* `gemini-free`
* `gpt-oss-120b-free`
* `mistral-small-free`
---
### 🌿 N1N
1. **Register:** [api.n1n.ai](https://api.n1n.ai)
2. **Copy API Key:** Click the "two pages" icon [here](https://api.n1n.ai/console/token).
3. **Choose Model:** [Go here](https://api.n1n.ai/console/personal).
4. **Setup in JanitorAI:** Bot Chat - Proxy - Create New Configuration - Save.
* **Model:** Your chosen model.
* **URL:** `https://api.n1n.ai/v1/chat/completions`
* **API KEY:** Your copied key.
#### 🍃 Free Models:
*(Just a few models, see the rest on the website. They are constantly changing.)*
*Paste these into the Model field (URL stays the same):*
* `claude-3-7-sonnet-latest`
* `deepseek-v3.2-thinking`
* `gemini-pro-latest-thinking`
* `claude-opus-4-5-20251101`
---
# ⛩️PROVIDERS WITHOUT DEEPSEEK
---
### 🌊 INFERENCE
1. **Register:** [inference.net](https://inference.net)
2. **Choose Model:** Select one from the list or copy from below.
3. **Get Key:** Click on `=` - then click on **API KEY**.
4. **Setup in JanitorAI:** Bot Chat - Purple Proxy Button - Proxy - Create New Configuration - Save.
5. **Fill as follows:**
* **Model:** Your chosen model (e.g., `mistralai/mistral-nemo-12b-instruct/fp-8`)
* **URL:** `https://api.inference.net/v1/chat/completions`
* **API KEY:** Your copied key.
#### 🍡 Free Models:
*(Just a few models, see the rest on the website. They are constantly changing.)*
*Paste these into the Model field (URL stays the same):*
* `google/gemma-3-27b-instruct/bf-16`
* `meta-llama/llama-3.2-11b-instruct/fp-16`
* `mistralai/mistral-nemo-12b-instruct/fp-8`
---
### 🎐 UPSTAGE
1. **Register:** [console.upstage.ai](https://console.upstage.ai/api-keys?api=chat)
2. **Copy API KEY** from the dashboard.
3. **Setup in JanitorAI:** Bot Chat - Purple Proxy Button - Proxy - Create New Configuration.
4. **Fill details:**
* **Model:** e.g., `solar-pro2-250710`
* **URL:** `https://api.upstage.ai/v1/chat/completions`
* **API KEY:** Your copied key.
#### 📜 Free Models:
*(Just a few models, see the rest on the website. They are constantly changing.)*
*Paste these into the Model field (URL stays the same):*
* `solar-mini-250422`
* `solar-pro-250422`
* `solar-pro2-250710`
---
### 🍃 MISTRAL
1. **Register:** [console.mistral.ai/upgrade/plans](https://console.mistral.ai/upgrade/plans)
2. **Plan:** Click on the **"Experiment"** tier.
3. **Copy Key:** Create and copy your key from the dashboard.
4. **Setup in JanitorAI:** Bot Chat - Proxy.
5. **Fill as follows:**
* **Model:** `mistral-medium-2505`
* **URL:** `https://api.mistral.ai/v1/chat/completions`
* **API Key:** Your copied key.
---
### 🎋 ARLIAI
1. **Register:** [arliai.com](https://www.arliai.com)
2. **Copy Key:** Create and copy your API key.
3. **Setup:**
* **Model:** `Gemma-3-27B-ArliAI-RPMax-v3`
* **URL:** `https://api.arliai.com/v1/chat/completions`
* **API Key:** Your copied key.
---
### 🕯️ PERPLEXITY LABS
1. **Register:** [labs.perplexity.ai](https://labs.perplexity.ai)
2. **Get Key:** Obtain the API key in Account Settings.
3. **Setup:**
* **Model:** `llama-3-70b-instruct`
* **URL:** `https://api.perplexity.ai/chat/completions`
* **API Key:** Your key.
---
### 🌪️ GROQ
1. **Register:** [console.groq.com](https://console.groq.com)
2. **Get Key:** Create key in the **API Keys** section.
3. **Setup:**
* **Model:** `llama3-70b-8192`
* **URL:** `https://api.groq.com/openai/v1/chat/completions`
* **API Key:** Your key.
---
### 🌑 DEEPINFRA
1. **Register:** [deepinfra.com](https://deepinfra.com)
2. **Get Key:** Dashboard - API Keys.
3. **Setup:**
* **Model:** `meta-llama/Llama-3-70b-instruct`
* **URL:** `https://api.deepinfra.com/v1/openai/chat/completions`
* **API Key:** Your key.
---
### 🔍 JINA
1. **Register:** [jina.ai](https://jina.ai/)
2. **Copy Key** from your dashboard.
3. **Setup:**
* **Model:** `reader-lm-1.5b`
* **URL:** `https://deepsearch.jina.ai/v1/chat/completions`
* **API Key:** Your copied key.
---
## 👘 CEREBRAS
1. **Register:** [cloud.cerebras.ai](https://cloud.cerebras.ai/)
2. **Get Key:** [API Keys Section](https://cloud.cerebras.ai/platform/org_xvdv6k8tv29mkytej4j4pp33/apikeys)
3. **Choose Model:** Select one from the list below.
4. **Setup in JanitorAI:** Bot Chat — Purple Proxy Button — Proxy — Create New Configuration — Save.
5. **Fill as follows:**
* **Model:** Your chosen model (e.g., `llama-3.3-70b`)
* **URL:** `https://api.cerebras.ai/v1/chat/completions`
* **API KEY:** Your copied key.
#### 🌸 High-Speed Models:
*(Full list: [Cerebras Docs](https://inference-docs.cerebras.ai/models/overview))*
*Paste these into the Model field:*
* `llama-3.3-70b`
* `llama3.1-8b`
* `qwen-3-32b`
* `qwen-3-235b-a22b-instruct-2507`
* `gpt-oss-120b`
* `zai-glm-4.7`
---
# 🌸CONNECTING GEMINI
(only official methods)
---
## 🩷 Google AI Studio
1. ✨ **Generate API key:** [**[Get your key here]**](https://aistudio.google.com/app/api-keys)
2. ✨ **Open Janitor** → Select your Bot → **Proxy** (top right corner).
3. ✨ **Fill in the details:**
> **Model:** `[Your model name]`
> **URL:** `https://generativelanguage.googleapis.com/v1beta/chat/completions`
> **API KEY:** `[Your copied key]`
[📋 Model Names](https://ai.google.dev/gemini-api/docs/models?hl=en#gemini-2.5-pro) | [📊 Quota Limits](https://console.cloud.google.com/apis/api/generativelanguage.googleapis.com/quotas?hl=en)
> 💡 Look for **"per quota free tier"**. Currently, they provide up to 20 messages per day.
**Free Tier Models:**
`gemini-2.5-flash-lite` | `gemini-3-flash-preview`
---
## 🐍 Personal URL Method (Colab)
> ⚠️ **Important:** VPN is required! Limits reset at 10:00 AM MSK.
1. 🌸 Create a Google account.
2. 🌸 Copy your API KEY: [[Link]](https://aistudio.google.com/apikey). Click "Set billed" (you can enter random data).
3. 🌸 Go to Google Colab: [[Open Script]](https://colab.research.google.com/drive/1xlqI0LeZLfbJRHOE92NrXbSVR1-SvQ7X)
Look for the ▶️ button and press it. Select the **Gemini** model.
### ⚙️ "Thinking" Configuration
*Determines if the bot "thinks" before answering (affects quality):*
* **Enable_thinking** * *Toggle ON* → OFF (Faster)
* *Toggle OFF* → ON (Higher quality)
* **Display_thinking** * Show bot's thoughts in chat? (ON - No / OFF - Yes)
**My Recommended Settings:**
Press ▶️, wait 1-3 minutes, and look for the link **"Running on..."** in the logs.
**Final Janitor Setup:**
* **Model name:** leave empty
* **URL:** link from Colab (Cloud)
* **API:** Your Gemini API key
---
## ☁️ CLOUD Setup (Anthropic)
1. ✨ Open Colab: [[Open]](https://colab.research.google.com/drive/1OYj_IUCu5Wc8T3Wtxnf0iRul17CPEF4K)
2. ✨ Select the model and press ▶️
3. 
4. ✨ Scroll down the log and find **"Running on: [link]"**.
5. ✨ Register Anthropic API: [[Here]](https://console.anthropic.com/settings/keys)
**Janitor Proxy Settings:**
> **Model:** The one you selected
> **URL:** Cloud link from Colab
> **API Key:** Key from your Anthropic dashboard
*Example of a finished setup:*
---
# 🌸 Gemini via Sophia (Lorebary) 🌸
---
This method uses **Sophia’s Lorebary** proxy to help bypass filters and provide a better experience for uncensored roleplay.
### 🔑 Step 1: Get Your Gemini API Key
1. ✨ **API Studio:** Go to [aistudio.google.com/apikey](https://aistudio.google.com/apikey).
2. ✨ **Create Key:** Click "Create API Key" and copy it.
3. ⚠️ **Pro Tip:** Use a "burner" Google account. While rare, using Gemini for NSFW content carries a risk of account action from Google.
---
### ⚙️ JanitorAI Configuration
Go to your **Bot** - **Proxy** (top right) -**Create New Configuration**.
> **Configuration Name:** `Gemini Sophia`
> **Model:** `Choose from list below`
> **URL:** `https://lorebary.sophiamccarty.com/aistudio`
> **API KEY:** `[Paste your Gemini API Key]`
---
### 💮 Sophia's Special Commands
To get the most out of this proxy, add these commands to the **end** of your **Custom Prompt** in API Settings:
> `<PREFILL_ON>`
> `<JAILBREAK=ON>`
> `<FORCETHINKING=ON>`
> `<AUTOPLOT=ON>`
> `<BETTERSPICE=ON>`
✨ **There may be several codes, up to 10 pieces.**
>[More codes can be found on Sophie herself](https://janitorai.com/external-link?to=https%3A%2F%2Florebary.sophiamccarty.com%2Fplugin-library)
---
### ⚠️ Troubleshooting
* ❌ **Error 500:** Internal Google error. Usually temporary—just wait a few minutes.
* ❌ **Error 429:** You have reached your daily message limit or rate limit.
* ❌ **Network Error:** Common during high traffic. Try again later or refresh.
* ❌**404 Error:** Double-check that your Proxy URL is exactly: `https://lorebary.sophiamccarty.com/aistudio`
---

## 🏮 TROUBLESHOOTING & COMMON ERRORS 🏮
---
### ⚠️ Network Error - Try Again
This occurs during server overload or unstable internet connection.
**What to do:**
* Ensure your internet is working properly.
* Switch between mobile data and Wi-Fi (or vice versa).
* **Disable VPN.**
* Verify that all Proxy data (URL, Model, Key) is entered correctly.
* Wait 5–10 minutes and try again.
* If the issue persists, switch to a different model or try another proxy method.
---
### ⚠️ Bot cuts off message / Incomplete text
This is a common generation issue.
**What to do:**
* Go to Generation Settings and **Disable Text Streaming**.
* Try lowering the **Temperature** or **Top P** settings.
---
### ⚠️ Not Enough Quota
You have reached your token or message limit.
**What to do:**
* Create a new account and generate a new API Key.
* Alternatively, wait for the daily reset if the service offers daily limits (like Rout.my or OpenRouter).
---
### ⚠️ Rate Limited / Failed to Fetch
Too many requests in a short time.
**What to do:**
* Refresh the page and wait. Double-check your Proxy URL and Model Name for any hidden spaces or typos.
---
### ⚠️ PROXY ERROR 401: No auth credentials
Wrong or expired API key.
**What to do:**
* Generate a fresh API key from the provider's dashboard and replace the old one in Janitor.
---
### ⚠️ PROXY ERROR 402: Insufficient credits
Daily limit reached for the current key.
**What to do:**
* Switch to a new account or wait for the reset (usually at 00:00 UTC).
---
### ⚠️ Bot is speaking Chinese or gibberish
The model is likely "hallucinating" due to high settings.
**What to do:**
* Lower the **Temperature** (try 0.7 - 0.9) or adjust **Top P**.
---
### ‼️ Gemini Specific Errors
*If you encounter a stuck or unresponsive error with Gemini models, try these fixes:*

**⚠️The bot doesn't write,although it seems to write**
**What to do:**
* Turn on streaming
---

⚠️ **An error occurred while processing your reguest (unk)**
---
**What to do:**

1. **Switch Projects:** If you have multiple projects (and keys), simply switch to a different one in your Janitor configurations.

2. **The Penalty Trick:** Go to your Generation Settings - set **Frequency/Presence Penalty** to **1** - and spam the "Send" button multiple times. Once the bot starts responding, set the penalty back to **0** and it should be back to normal.

3. **Refresh the Key:** Generate a brand new API Key and replace the old one.

4. **Wait for Reset:** If none of the above work, you have likely reached your message limit. You will need to wait for the daily reset.
---

## ✨Feedback

Perhaps in the future this guide will be updated and some proxies will be removed. If you notice that the proxy no longer offers a free plan, please write [Google Form](https://forms.gle/ZxmATSPgwoTgy58XA)