Will It Fit?

Inspiration

What it does

How we built it

Challenges we ran into

Inspiration

I wanted to make it easier for people to run AI models on their own computers. Choosing a model can be confusing because its memory requirements may exceed what a laptop or GPU can handle. I wanted to help users make an informed decision instead of just seeing an error.

What it does

Will It Fit? estimates whether an AI model can run on a user's hardware. It considers model size, precision, context length, GPU memory, RAM, and storage. It recommends possible fixes, such as using a smaller or quantized model, and compares suitable cloud GPU options with local hardware costs.

How we built it

I used the Devpost Learn Skill Pack with an AI coding agent. We followed a planning-first workflow, creating the scope, product requirements, and technical specification before implementing the application in four slices. We used JavaScript, HTML, and CSS, added automated tests, and checked the application in a browser.

Challenges we ran into

Estimating memory usage accurately was challenging because model weights are only part of the requirement. Context memory and runtime overhead also matter. We also had to account for system RAM limits and real-world GPU pricing, including unavailable retail listings. We refined the recommendations and tested the application against different hardware configurations.

Accomplishments that we're proud of

The main accomplishment is turning a confusing hardware compatibility problem into a practical decision-making tool. Instead of only showing whether a model fits, Will It Fit? suggests possible solutions and compares local upgrades with cloud GPU rental costs.

What we learned

We learned the importance of planning before coding, breaking work into smaller steps, and testing AI-generated code. We also learned to verify calculations, document assumptions, and review recommendations rather than trusting generated code without checking it.

What's next for Will It Fit?

Future improvements could include a searchable model catalogue, more model architectures, updated hardware prices, and additional cloud providers. We could also improve estimates using real inference benchmarks and make it easier for users to compare performance, cost, and privacy.

Accomplishments that we're proud of

What we learned

What's next for Will It Fit?

Built With

• ai
• claude
• copy
• css
• git
• html
• javascript
• node.js