Skip to content

SlickGPT is a light-weight "use-your-own-API-key" web client for the OpenAI API written in Svelte. It offers GPT-4 integration, a userless share feature and other superpowers.

License

Notifications You must be signed in to change notification settings

ShipBit/slickgpt

Repository files navigation

Contributors Forks Stargazers Issues MIT License LinkedIn

Built With

Svelte SvelteKit Skeleton


Table of Contents
  1. About The Project
  2. Features
  3. Comparison with ChatGPT
  4. Roadmap
  5. Development
  6. License
  7. Contact

New: SlickGPT Pro (use without API key)

SlickGPT now offers an optional "Pro" plan for a fixed price. If you choose this, you can use our Azure cloud infrastructure to access all the OpenAI models and settings without worrying about having your own API key or usage limits. Yes, it's unlimited usage.

We currently offer these models in SlickGPT Pro:

  • GPT-3.5 Turbo
  • GPT-4 Turbo
  • GPT-4o

There are no paywalled extra features in SlickGPT for Pro users. We don't log any data - the only thing we know is how many tokens were spent. It's just a way to support the project, get rid of the hassle of managing your own API key and to pay a fixed price for unlimited AI usage.

If you're really into AI, you can also combine it with our other project, Wingman AI, and get Wingman AI Ultimate (which includes SlickGPT Pro) for a reduced price.

About the project

SlickGPT

SlickGPT is a light-weight "use-your-own-API-key" (or optional: subscription-based) web client for OpenAI-compatible APIs written in Svelte. It offers them a very fancy user interface with a rich feature set like managing a local chat history (in your browser's IndexedDb), a userless "Share" function for chats, a prominent context editor, and token cost calculation and distribution. You can run the client locally, host it yourself or use our public instance.

Feature Overview

  • Fancy User Interface
    Interact with an elegant and cleanUI that makes chatting fun and explains a lot of the heavy stuff to users.
  • Mobile first and fully responsive
    SlickGPT is always at your hand. Literally, on your phone.
  • Local History
    Your chat history is stored in your browser's IndexedDb and never leaves your machine (until you actively decide to share chats).
  • Delete and edit messages
    Just like you are used to from the OpenAI client.
  • Auto-suggest chat titles
    SlickGPT can suggest a title for your chat based on the first message you send.
  • Share Chats
    Generate shareable links for your chats with 3 clicks to collaborate and brainstorm ideas with your team or friends. Without any authentication.
  • Set Context
    Provide your own context to guide the AI model while giving responses, ensuring relevant and focused outcomes.
  • Cost Calculation
    Keep track of and control your API usage costs with an in-built token cost calculator for your chats.
  • Full Configuration Power
    Customize OpenAI model settings to tailor the experience to your liking.
  • No Registration or Login required
    Create and even share chats without the need for a login or registration - Simple, fast, and efficient. Note that you'll have to register/login if you want to use the SlickGPT Pro subscription.
  • Use your own API Key
    Keep your API calls private and secure by using your own OpenAI API key or use our cloud infrastructure if you don't want to have an API key or prefer a fixed price over a "pay-per-use" model.
  • Host Yourself / Run Everywhere
    Run the instance locally on your machine or host it on your preferred cloud provider.

SlickGPT currently supports these models:

  • OpenAI:

    • GPT-3.5 Turbo
    • GPT-4 Turbo
    • GPT-4o
  • Mistral:

    • Mistral large (latest)
  • Meta (via Groq):

    • Llama 3 8B 8192
    • Llama 3 70B 8192

Theoretically, every LLM that understands OpenAI's conversation schema should work with SlickGPT. If you want to use a model that is not listed here, please open an issue or contact us. If you self-host SlickGPT, you can also connect it to a local model running in Ollama, LMStudio etc.

See a YouTube video for a quick feature overview (from an older version):

YouTube video of SlickGPT

[ back to top ]

In more detail

Local chat history

Your chat history is stored in the IndexedDb of your browser and never leaves your machine (unless you actively decide to share a chat.

History of chats

Share Chats

Share chats with 3 clicks to collaborate and brainstorm ideas with your team or friends. No authentication or registration necessary. You can also unshare or update shared chats later. Recipients can import shared chats with a single click and continue your conversation in their own local chat instance.

Share chats

Change advanced settings

SlickGPT allows you to customize all parameters the OpenAI API offers to your liking. You can change the model, temperature, top_p, and frequency_penalty parameters to get the best results. You can also set the the chat context (or "system" message) to guide the AI model and shape its behaviour and answers and even use "stop sequences" to limit the answers you get.

Settings

Detailed token calculation

SlickGPT provides an in-built token cost calculator for your chats. You can see the cost of your chat in real-time as you type and receive answers. This allows you to keep track of and control your API usage. SlickGPT also has a posh pie chart that visualizes the cost distribution within a chat.

Token calculation

Change context

Do you know what "context" is when talking to ChatGPT? Most users don't and that's why SlickGPT displays and explains the concept very clearly with explanations and examples when you start a new chat. This way, one of the most powerful features just becomes "part of the chat".

Set context

Switch between light and dark mode

SlickGPT offers a light and dark theme. You can switch between the two modes by clicking the toggle button in the top right corner and once you do, it will remember your choice.

Switch theme

[ back to top ]

Comparison with ChatGPT

While ChatGPT offers a great conversational AI experience, it lacks some features many users desire.

Feature SlickGPT ChatGPT Plus
Use different models (✅ OpenAI models only)
Chat history
Highlight / Format code
Use Your Own API Key
Edit and branch messages
Stop generation
Vision
Share and continue chats (✅) (can't import)
Host yourself / Run locally
Set context / system message
Change model configuration
No Login Required
Token Cost Calculation
Mobile-first design
Unlimited use subscription 5,99 €/m 18,00 €/m

As you can see, SlickGPT includes several additional features that cater to a broad range of user needs, such as hosting the service yourself, using a local chat history, sharing chats, and customizing the chat configurations. Moreover, SlickGPT does not require a login, allowing for faster and more streamlined access. These added functionalities make SlickGPT a more versatile and convenient solution for users seeking an advanced ChatGPT experience.

[ back to top ]

Roadmap ideas

  • Integrate with Wingman AI
  • More documentation
  • add Vision & image generation
  • add file uploads
  • update to Svelte 5 and newer dependencies
  • i18n
    • English
    • German

See the open issues for a full list of proposed features and known issues.

[ back to top ]

Development

SlickGPT uses Svelte, SvelteKit and Skeleton UI as frontend foundation.

The secret sauce of SlickGPT is that almost everything is stored in the IndexedDb of the browser. One case where it leaves this boundary is, of course, when a user communicates with the OpenAI API. While questions are simple POST requests, responses are sent back to the client using the native EventSource API and rendered as a "live stream" just like the ChatGPT client does.

The other case is when users share their chats with others using the "userless share" feature. Then SlickGPT dumps the appropriate IndexedDb object in a simple Firebase Realtime Database and tags it with a random updateToken. A serverless (Edge) function sends this token back to the client where it is stored in the IndexedDb next to the original chat. The token can be used later to update or unshare chats.

SlickGPT uses Svelte endpoints as "proxy" between the client and the external APIs. This way, the client does not need to know any Firebase secrets at all and you have a great starting point to implement more complex "backend" logic if needed.

Set environment Variables

Copy the content of .env.example to a new file called .env and populate the variables with your Firebase data. You can get this firebaseConfig from your project settings in the Firebase console.

While developing, Vite will be your "server" and run the endpoints functions on your local machine. It has access to your environment variables utilizing SvelteKits $env/static/private import.

For a production environment, the endpoints are preconfigured to run as edge functions on Vercel. For this to work, you simply have to provide Vercel with the required environment variables once you go live with your instance. You can paste the content of your .env file when you import your project to Vercel or afterwards in the project settings or using the Vercel CLI.

Start the client

Once cloned, run the usual SvelteKit commands in your root directory:

npm i # or yarn, pnpm etc.
npm run dev -- --open

That's it! You should now be able to develop with all the hot reloading goodness you know and love from Svelte / Vite.

Can I run this without Firebase?

Certainly. The quickest and easiest way would be to disable the "Share Chat" feature in SlickGPT in case you don't need it. We have an open issue for that with good starting points.

The other (and better) way would be to use any other database where you can dump SlickGPTs serialized Chat objects. This should be an easy task because the client never directly talks to Firebase as described above. SlickGPT doesn't utilize any Firebase-specific features. As long as your database can somehow save and load a chat based using a slug as key, you should be fine. The only file you'd have to edit is the share endpoint.

Can I host this on other providers than Vercel?

There's an open issue for that with some additionale infos regarding Netlify. We haven't tested other providers yet but it should be fine as long as your provider can run SvelteKit apps and serverless functions. We already saw a fork utilizing Railway.

You could also let the client access the Firebase and OpenAI API directly to get rid of the serverless function requirement entirely.

How do I use another Skeleton Theme?

Simply follow the instructions in the official Skeleton UI documentation. The code you have to change is in SlickGPTs +layout.svelte file.

Contribution Guidelines

Contributions are always welcome! Feel free to open a pull request, report bugs, or submit feature requests. Please follow the Code of Conduct and Contribution Guidelines when participating in this project.

Main Contributors

A portrait of Simon Hopstätter
Simon Hopstätter

A portrait of Timo Korinth
Timo Korinth

[ back to top ]

License

This project is licensed under the MIT License.

Contact

SlickGPT was initially created by Simon Hopstätter and Timo Korinth at Shipbit.

If you have any questions, feel free to contact us on our Discord server.

Logo

[ back to top ]

About

SlickGPT is a light-weight "use-your-own-API-key" web client for the OpenAI API written in Svelte. It offers GPT-4 integration, a userless share feature and other superpowers.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published