A model in
your tab.

This is not a video of a chatbot. It is a real 0.5B language model running on your own GPU — about 265 MB, once. It knows Parth's projects, skills and story; flip Web on and general questions can look things up on Wikipedia. The conversation never leaves this tab unless you ask it to.

CHECKING WEBGPU~265 MB once4,096 tokensNo server

The machine

The machine
LocalQwen2.5 · 0.5B4K contextIdle

Ask me about Parth.

A real language model, running on your GPU. It knows Parth's projects, skills and story, and loads only when you ask.

~265 MB one-time download · runs on your GPU · nothing you type leaves this tab

Nothing is downloading until you wake the machine.

What runs

Qwen2.5 0.5B, 4-bit, through WebLLM and WebGPU in a Web Worker. Under a gigabyte of GPU memory and a 4,096-token context.

What leaves

Nothing, once the model is cached. No analytics, no cookies, no chat storage. With Web on, only a general question is sent to Wikipedia.

What if it can't

Without WebGPU, the machine answers instantly from the same notes. The rest of the site never depends on it.

What actually
happens.

Four steps, no server in the loop. The only network requests are the model download and, if you turn it on, the Wikipedia lookups Web mode makes for you.

01

You press wake

The site checks for WebGPU. If it is available, it starts a dedicated Web Worker and begins the one-time model download.

02

The model loads

Qwen2.5-0.5B-Instruct-q4f16_1 is fetched from a public CDN, quantised to 4-bit, and loaded into GPU memory through WebLLM. The browser caches the weights.

03

It reads the question

Plain code works out what you're asking — which project, which detail, whether it follows on from the last answer — and hands the model the two or three fact cards that answer it. A small model writes best from a short, focused brief.

04

The answer streams

Tokens are generated on your GPU and streamed into the chat. Sources are shown as chips. Reload the page and the conversation is gone.

What it can do

  • Answer questions about all seven chapters, the side projects, Parth's education, internship, skills and how to reach him.
  • Follow a conversation — “tell me more”, “what about its results?” — without losing the thread.
  • Answer general questions from what the model knows, or from Wikipedia with Web on, citing the page.
  • Show where each answer came from as source chips.
  • Say it doesn't know instead of inventing a fact about Parth.
  • Run offline once the model weights are cached and Web is off.
  • Answer instantly from the same facts when WebGPU is unavailable.

What it cannot do

  • Browse the internet — until you turn Web mode on, which only ever talks to Wikipedia.
  • Remember a conversation after you reload the tab.
  • Match a large cloud model on open-ended reasoning.
  • See images, hear audio or run tools.
  • Send your conversation anywhere — Web mode sends only the search query, and it is off unless you turn it on.

The specification

Small enough to load quickly, large enough to answer a question about the work.

Read the colophon
Model
Qwen2.5-0.5B-Instruct-q4f16_1-MLC
Download
~265 MB
GPU memory
~945 MB
Context
4,096 tokens
Runtime
WebLLM · WebGPU · Web Worker
Licence
Apache 2.0