How Will Open-Source AI Make Money?

Last week, we spent a lot of time talking about Meta Platforms’ new open-source Llama 3 models. However, the one question that still remains unanswered is how Meta and other developers of open-source AI software will make money from it—an uncertainty that likely contributed to Meta’s stock drop on Thursday.

Take Paris-based startup Mistral, for example, currently in talks to raise at a $5 billion valuation just months after raising at a $2 billion price. The large language model developer so far has released a number of widely-acclaimed LLMs, but most are free open-source versions, with a few paid ones that have been optimized for better performance. 

In the long run, the startup hopes that customers will first experiment with its smaller, open-source models and eventually move onto its larger, paid ones. The startup plans to make money from these larger commercial models as well as a yet-to-be-released application programming interface that will allow customers to tweak, or finetune, its models, said a person with knowledge of the startup’s thinking.