arXivMax turns research papers into animated video explainers. Add “max” after “arxiv” in any arXiv URL, press Generate, and watch. Built for researchers and engineers who want a fast first pass before deciding where to read deeply. Works great as a passive listen for building intuition on the latest research in AI, Physics, and Math.

How did GPT-5.6 change the ambition or scope of what you shipped?

I actually built this product before GPT-5.6 released, and I struggled to get the codegen for the Manim visuals to work well. I swapped in GPT-5.6 Sol the day it released publicly and the quality 10x'ed literally with a single line change! 5.6 is much better at visuals and codegen, and cost-efficient, than any other models out there.

Hey Product Hunt 👋 I bookmark arXiv papers faster than I can read them, so I built arXivMax. arXivMax turns arXiv papers into animated, intuition-first video explainers. Think 3Blue1Brown-style visuals, generated for the paper you’re trying to understand. This isn’t meant to replace reading papers. I use it as a fast, passive first pass while coding, commuting, or tokenmaxxing, then go back and revisit the paper if I want to dive deeper. How it works: 1. Open any arXiv URL 2. Append “max” after “arxiv”: eg arxiv.org/abs/2006.11239 → arxivmax.org/abs/2006.11239 3. Press Generate 4. Watch the paper The full demo above explains Denoising Diffusion Probabilistic Models: the original 2020 DDPM paper by Jonathan Ho, Ajay Jain, and Pieter Abbeel. Diffusion models now power image and video generation and are spreading into audio, robotics, weather forecasting, and drug discovery! Right now, arXivMax is entirely free, and I don’t have plans to monetize it anytime soon. Try it, break it, and tell me how I can make it better. - Richard

How did GPT-5.6 change the ambition or scope of what you shipped?

arXivMax

Turn any arXiv paper into a video explainer

Turn any arXiv paper into a video explainer