As AI models continue to grow in size and complexity, a burning question has emerged: how do hobbyists and professionals alike afford to run these massive models locally?
I’m guilty of being curious, and I’m not alone. A fellow Redditor recently asked the same question, sparking a wave of responses from those who’ve ventured into the world of large-scale AI modeling.
## The Cost of Running Large Models
For those of us who dabble in model fine-tuning as a hobby, the cost can be daunting. I’m lucky to have a decent setup, with a 3090ti GPU and 64GB of RAM, but I mostly stick to models under the 10b parameter range. Anything larger, and I’d need to fork out for cloud services or invest in some serious hardware upgrades.
## The Professionals: How Do They Afford It?
So, what about those who run massive models as part of their daily job? Are they simply paying for cloud services or do they have access to stacks of A100/H100s? It’s hard not to wonder how they afford it, especially when you’re struggling to make ends meet in your own industry.
## My Story: A CGI Generalist Turned AI Enthusiast
As a CGI generalist and 3D character artist, I’ve seen my industry take a huge hit. With extra free time on my hands, I’ve turned to learning AI model fine-tuning and data formatting. It’s not a full-blown career change (yet!), but I’m drawn to the excitement of new tech, even if it’s disrupting my current field.
## The Future of AI Modeling: Accessibility and Affordability
The conversation raises important questions about the accessibility and affordability of AI modeling. As models continue to grow in size and complexity, will hobbyists be priced out of the game? Or will cloud services and hardware advancements make it possible for everyone to participate?
What’s your take on the cost of running large AI models? Share your experiences and insights in the comments below!