Electron plasma

I need an application that replicates the end-user experience of Google’s Gemini while letting me supply my own API key. Once the key is entered, every feature must unlock automatically—no hard-coded credentials.

Core functionality
• Full AI feature suite: chat-based text generation with streaming responses, image generation, and robust LaTeX rendering (MathJax or KaTeX are fine as long as formulas display crisply across devices).
• Clean, responsive interface that mirrors Gemini’s conversational flow, shows conversation history, and supports markdown, code blocks, and inline math.
• Simple “Settings” panel where I paste or update the key; all calls should route through that stored key.
• Modular codebase so I can later swap models (OpenAI, Gemini, or any endpoint using the same JSON schema).
• Basic usage stats: token count, request latency, and daily call total shown unobtrusively in the UI.

Deliverables
1. Source code (React + Node.js preferred, but I’m flexible if the stack is well documented).
2. README with setup, build, and deploy instructions so I can run it locally or push to Vercel/Netlify.
3. Brief video or screenshots confirming LaTeX equations, text chat, and an image generation prompt all work through my supplied key.

Acceptance criteria
– After I enter an API key and refresh, every feature above functions without additional configuration.
– LaTeX renders correctly in chat history on both desktop and mobile breakpoints.
– Image prompts return URLs or inline previews.

Keep the UI minimal—Gemini-style but not a pixel-perfect clone—so performance stays snappy. Once the base is solid, I’ll open a second phase for extras like multi-account support and 2-factor auth.


I don't need exact copy, i want similar and more advanced