ChatKit

ChatKit is a refined ChatGPT interface that works with your own OpenAI, Gemini, Azure or custom API key. It adds prompt templates, chat bookmarks, file drops, web search, URL fetching, voice input and text-to-speech, with a one-time payment option and weekly feature updates.

ChatKit is a front-end layer for ChatGPT that replaces the stock interface with a more configurable workspace. The pitch is simple: keep the models you already trust, but swap the surrounding chat window for something with prompt templates, saved conversations and adjustable generation settings. Instead of a single locked provider, ChatKit accepts your own API key and can point at OpenAI, Gemini, Azure or custom model endpoints, so the same client can serve several backends without switching apps. In practice the tool behaves like a power-user console. You paste an API key, which the site states is stored locally, then pick a model and set a temperature of your choice. Prompt templates save frequently used instructions so they do not have to be retyped, and chat bookmarks keep important threads within reach. File drops into the conversation are supported, web search and URL fetching bring outside material into the chat, and text-to-speech plus voice input cover hands-free reading and dictation. Users who would rather not manage a key can run on ChatKit credits through a premium upgrade. It suits people who already live inside ChatGPT and want more control over how requests are made: developers testing prompts across providers, researchers collecting sources through URL fetches, and anyone who wants saved snippets and bookmarks without building their own client. Native desktop chat apps such as ChatWise occupy adjacent ground, while ChatKit's differentiator is the browser-based refined UI paired with a one-time payment rather than an ongoing per-seat license. A few caveats are worth flagging. The model list and feature set depend on what each provider exposes, so custom endpoints may not support every control. Much of the value presumes comfort with API keys and model parameters, which puts it closer to intermediate than beginner territory. And because features ship weekly, the interface can shift under you between sessions.