💭
Memory
Remembers who's in the room, their personalities and what you talked about recently.
Open source · MIT · Powered by Flatkey
Your own AI chatbot,
in every chat.
Mochi remembers the people she talks to, answers from your docs, looks things up and reads images. Run as many bots as you like on Discord, Telegram, Slack, QQ, LINE and your website, all from one dashboard.
One Flatkey key powers every model Mochi uses. Mainstream models at least 20% below official pricing.
Runs onDiscordTelegramSlackQQLINEWebsite
# hangout● online
Everything a great chatbot needs
Turn each ability on or off per bot, and pick the model behind it.
📚
Knowledge base
Add docs, FAQs or notes and get a Q&A bot that answers from them.
🔎
Web search
Looks up facts and recent news instead of guessing.
🖼️
Reads images
Understands screenshots, photos and memes people send.
🤖
Multiple bots
Each bot has its own persona, description, channels and budget.
💬
6 channels
Discord, Telegram, Slack, QQ, LINE and a widget for your website.
🌏
4 languages
Dashboard in English, 中文, 日本語 and Português.
🪙
Cost control
Switch abilities off, choose cheaper models and watch spending per bot.
Live in three steps
- 1
Get a Flatkey key
Create an API key on Flatkey and top up your balance. It powers chat, vision, search and memory.
Get your Flatkey key → - 2
Deploy Mochi
Run it with Docker, on Sealos, on any VPS or on your own computer. It needs an always-on server.
Deploy guide (Docker, Sealos) → - 3
Connect and chat
Paste your key in the dashboard, set a persona, add knowledge and connect your channels.
git clone https://github.com/flatkey-ai/Mochi-ai-chatbot
cd Mochi-ai-chatbot && npm install
cp .env.example .env && npm start
# open http://localhost:3000/dashboardWhy Mochi runs on Flatkey
Official models. Better prices.
A chatbot needs several kinds of models. Flatkey gives you all of them behind one key.
Create your Flatkey keyPowered by flatkey
💸 Save money
Mainstream models like GPT, Claude, DeepSeek, Kimi and GLM are at least 20% below official pricing. Same official models, same quality. Failed calls cost $0.
⏱️ Save time
Chat, vision, web search and a low-cost model for memory. Going direct means one account and invoice per provider. Flatkey is one key, one balance and one bill for 100+ models.
🔁 Switch freely
Change the model behind any ability from the dashboard in one click. No new signup, no code change.
A dashboard anyone can use
What will your bot be?
🎮
Community companion
A friendly regular for your Discord or Telegram group who knows everyone by name.
🛟
Support Q&A bot
Answers customer questions from your docs on your website, LINE or Slack, around the clock.
📖
Study buddy
Explains, quizzes and looks things up for a class or study group on QQ or Discord.
Mochi AI chatbot FAQ
Is Mochi free?
Mochi is open source under the MIT license. You pay only for the model usage your bots make, from your Flatkey balance. Turn abilities off or pick lower-cost models to spend less.
Why only Flatkey and not my own OpenAI key?
A chatbot uses chat, vision, search and memory models from different providers. Flatkey puts 100+ official models behind one key and one bill, with mainstream models at least 20% below official pricing. That keeps Mochi simple to set up and to run.
Do I need to code?
No. You run one command (or a Docker container), then do everything else in the dashboard: personas, knowledge, channels and models.
Where is my data stored?
On your own server, in a local database file. Mochi sends messages to Flatkey only to get model replies.
Can I host it on Vercel or another serverless platform?
No. Bots keep live connections to Discord, Telegram and others, so Mochi needs an always-on process and disk. Use Docker, Sealos, a VPS or your own computer.
Which models can I use?
Any model in the Flatkey catalog, including GPT, Claude, Gemini, DeepSeek, Kimi and GLM. Mochi starts with a fast, low-cost DeepSeek model and you can change it any time.
Bring Mochi to your community today
Get a Flatkey key, deploy Mochi and say hi.