infersize

infersize

A browser tool that estimates the GPU memory an LLM needs, compares GPUs, sizes speculative decoding latency and estimates cloud cost per million tokens, before you rent or buy hardware.

Made by prajwalpujari16 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: infersize is a browser tool for planning LLM inference hardware before you commit money. It estimates the video memory a model needs, compares several GPUs in one matrix, sizes latency for speculative decoding and estimates cloud cost per 1M tokens. It is a static page, so nothing runs on a server.

Who it suits: It suits people who must choose GPUs or cloud instances for serving a language model and want numbers before they decide.

Title: Who might use this tool
Industry | Who uses it | How they use it | Example
ML platform teams | Infrastructure engineers | Size the GPUs a model needs before ordering | Check whether a 7B model fits a card
Startups | Founders | Estimate what serving a model would cost | Read the cloud cost per 1M tokens
AI research | Researchers | Reason about speculative decoding latency | Compare setups in the latency sizer

If you need measured benchmarks from real hardware, you can skip it, since the page models performance analytically.

Where to use it: You can use it online or open its files on your own computer.

• Web browser: the Space's page on Hugging Face, with nothing to install
• Windows, macOS or Linux: the downloaded files, opened in a browser
• Python: a PyPI package of the same name, named in the readme

Title: What the tool is good for
Use case | Who | What they get back
Size memory for a model | Engineers | An estimate of the VRAM the model needs
Compare GPUs | Buyers | A matrix that lines several GPUs up side by side
Budget cloud serving | Founders | A cost estimate per 1M tokens

How to set it up: The hosted page needs nothing. To get the files you need Python and pip.

• Python, with pip
• Any web browser

1. Install the huggingface_hub package.
2. Download the Space's files into a folder.
3. Move into that folder.
4. Open index.html in a web browser.

pip install -U huggingface_hub
hf download prajwalpujari16/infersize --repo-type space --local-dir infersize
cd infersize

The readme also names a PyPI package, which installs with this command.

pip install infersize

How to use it: Open the Space online, or open index.html from your downloaded folder. Then work through the four tools on the page, starting with the memory estimate.

Title: The four tools on the page
Tool | What it does
VRAM sizing | Estimates the video memory an inference setup needs
Multi-GPU comparison matrix | Lines up several GPUs side by side
Speculative decoding latency sizer | Estimates the latency of speculative decoding
Cloud TCO estimator | Estimates cloud cost in dollars per 1M tokens

Example: You pick Llama 3.1 8B, read how much video memory it needs, and check which GPUs in the comparison matrix can hold it. Then you open the cloud estimator to see the cost per 1M tokens.

For newcomers and experts: If you are new to this, start with one model and read only the memory estimate. VRAM is the main limit on which GPUs can run a model.

If you know the field, the page does analytical modeling, not measurement, so treat its numbers as pre-flight estimates. It names Llama 3.1 8B, Qwen2.5 7B Instruct and DeepSeek V2.5 as example models.

Choose a model
The page sizes VRAM analytically
Compare GPUs and decoding latency
Read the cloud cost per 1M tokens

Good to know: The estimates come from analytical models, so check them against your own tests.

• Static page: it is plain HTML and CSS, with no server code
• Estimates: the readme calls the sizing pre-flight analysis
• License: apache-2
• Cost to run: none named
• Data it reaches: nothing beyond its own files is named in the notes
• Not for: measured benchmarks of real hardware

On Hugging Face, 13 November 2026

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

Links

• Space: https://huggingface.co/spaces/prajwalpujari16/infersize
• Files: https://huggingface.co/spaces/prajwalpujari16/infersize/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.