According to this recent study from the Linux Foundation, the growth in interest and adoption of Open Source AI (OSAI) has been nothing short of phenomenal. Of course, one major driving force has been costs. The report estimates savings of 3.5X versus proprietary software. But we think that although cost may be an initial driving factor, there are other reasons why Open Source should be a preferred alternative.
Check out the study at this link.
Since the advent of DeepSeek R1, there’s no performance reason why not to select an Open Source model, even if you don’t want the fairly loose restrictions of Meta’s model, or if, like some you are philosophically opposed to potential contro by an American government or even just another American billionaire. Granted that DeepSeek comes from a country which is known for cyber-espionage, but we are relatively sure that once downloaded and run on its own hosting platform, there have been no indications of Chinese government control or interference.
But if these two models, which have excelled in their performance, are acceptable, there are probably and literally a million others to choose from.
So with cost and performance out of the picture, there is the question of having the horsepower to run these models. But even in that case, there are compression routines that would allow you to run a decent version of Deep Seek on hardware worth less than 10,000 dollars. Or at least for early experimentation, there are hosted version available in Canada at little cost.