Pluto 124M Chat

Pluto 124M Chat

A small chat model that runs entirely in your browser through onnxruntime-web, on WebGPU or WebAssembly, with no server doing the inference. For people who want a private, local chat demo.

Made by xxparthparekhxx on Hugging Face. These notes are Mindket's, written from its Space page and README; the app's own files and license are on Hugging Face.

About

What it is: This Space runs Pluto, a small chat model trained from scratch, inside your browser. It uses onnxruntime-web, with WebGPU where that is faster and multi-threaded WebAssembly otherwise. No server does the inference, so your messages are handled on your own device.

Who it suits: People who want to try a local chat model with nothing to install, and developers who want a working example of a model running in the browser.

Industry | Who uses it | How they use it | Example
AI development | Developers | Study an in-browser model setup | Read the static page and its ort folder
Education | Students | See what a small chat model can and cannot do | Open the Space and ask a short question
Privacy-minded work | Everyday users | Chat with a model that stays on the device | Try a prompt with no server involved

People who need long conversations or strong answers can skip it, since the model is small and its memory is limited.

Where to use it:

• A browser: the Space runs on Hugging Face with nothing to install
• Windows, macOS or Linux: open the downloaded index.html in a web browser

Use case | Who | What they get back
Local chat | Everyday users | Replies generated on your own device
Study the page | Developers | A static page that runs an ONNX model

How to set it up:

• Python with pip on Windows, macOS or Linux, only to download the files
• A web browser, ideally one with WebGPU
• Room in the browser cache for the model file, which downloads once

1. Install the Hugging Face hub package.
2. Download the Space's files into a folder.
3. Move into that folder.
4. Open index.html with any browser.

pip install -U huggingface_hub
hf download xxparthparekhxx/pluto-chat --repo-type space --local-dir pluto-chat
cd pluto-chat

How to use it: Open the Space page, or the local index.html, and type a message. The page formats your prompt in the chat style the model was fine-tuned on. Once a chat runs past the model's 1,024-token window, the page discards its earliest messages.

Example: You type a short question into the page. The model answers on your device, and the first load takes longer because the model file is fetched and then kept in the browser cache.

For newcomers and experts: If you are new, open the Space and ask something short. Keep expectations modest, because this is a small model of about 124 million parameters.

For experts, the page loads an fp16 ONNX export of about 243 MB with an explicit KV cache. The author compared it with the PyTorch model on 20 thousand held-out FineWeb-Edu tokens and reports that the two agree on the next token 99.8 percent of the time.

You type a message
The page builds the chat prompt
onnxruntime-web runs the model locally
The reply appears in the page

Good to know:

• License: apache-2
• Cost to run: none named
• Data it reaches: the model file it downloads once, while your messages stay on the device
• Not for: long conversations, since the context is limited to 1,024 tokens

On Hugging Face, 13 November 2026

• Likes: 1
• Built with: static
• License: apache-2.0

Links

• Space: https://huggingface.co/spaces/xxparthparekhxx/pluto-chat
• Files: https://huggingface.co/spaces/xxparthparekhxx/pluto-chat/tree/main

Try it

Open the Space in your browser, or copy it to your own Hugging Face account from its page. Nothing in this kit runs by itself.