The a16z Show · Thursday, August 6, 2026
Simon Mo recalls the 2023 release of Mistral Group's first model, which was distributed via a torrent link, highlighting the immediate need for GPU infrastructure to achieve efficiency. He notes that even then, users had to figure out how to run models on GPUs to make them faster or efficient for tasks.
“So, and, and even up to today, if you look at the K3 model release, is a whole partnership and a drive to make sure that the model is, once the model is released, because it's just a few terabytes of files sitting on the internet, that people are actually going to use it really, really well.”
“Even from the beginning of 2023, 2024, if you remember when Mistral, Mistral Group, dropped their first model, they just dropped a torrent, linked for PT, and then everybody's like, wow, you have to run on GPU to make it anything faster or efficient for any translation or any task.”