Have a project in mind?Have a project in mind? We design and build for the web.Start a project

Featured

Added on Aug 4, 2026

Serverless GPU Infrastructure for Real-Time AI

Serverless GPU Infrastructure for Real-Time AI

Serverless GPU Infrastructure for Real-Time AI

Share this project

About

Cerebrium runs voice agents, video models and LLMs on serverless GPUs, with sub-second cold starts, pay-per-second pricing and no Kubernetes. Its site, by Mathis Biabiany and KOKI-KIKO, opens on "Real-time AI infrastructure that scales with you" over a dark backdrop of glowing magenta ribbons, which later wrap a large translucent sphere behind sections on security and compliance. Lighter sections cover cold starts, elastic GPU scaling and observability, along with example builds and blog posts. It's Three.js on Astro with particles and simulation, and the page weighs under 1 MB.

What it is

SaaS/tech AI Geometric

Look

Minimal Sci-fi Pink Dark

How it works

Background Scroll-driven Drag/orbit

How it's built

WebGL2 Lenis Astro Custom shaders 3D models Particles Simulation 0.9 MB

Keywords

gpu voice agent ribbon sphere infrastructure serverless llm observability

Submitted by

mesh3d

Share this project

© All content, screenshots, and screen recordings are the property of their respective owners.

Similar Three.js experiences

Sit your work next to test, Abeto, and the rest.

We're always looking to feature fresh, creative work that pushes what's possible on the web.