Nah it’s the opposite. At the moment there’s a lot of low hanging fruit, AI companies can make money from the difference between GPT4 and free models.
Once we get free models that are as good as GPT4, while there will still be a market for a “GPT5”, the current applications will experience serious price competition.
It doesn’t matter, if you can run Llama3/4 on commodity hardware is what I’m saying.
If I can fulfill my objectives with a commodity model with GPT-4 capabilities then I’m not going to pay any $$$ for GPT-5 because my needs are already met. So OpenAI will lose customers at the same time as gaining them.
The price at the top end will go up, but the price at the bottom end (which is currently the top end) will fall.
I think Meta is the last major player who's only releasing free generative AI tools, so with no incentive to kneecap their best free stuff to avoid cannibalizing their own products. AI companies will have to build actual products on top of their models to produce more value, so I want to see if Meta will release whole free products to keep the pressure on. Maybe "Segment Anything" is already a step in that direction.
Comments
Nah it’s the opposite. At the moment there’s a lot of low hanging fruit, AI companies can make money from the difference between GPT4 and free models.
Once we get free models that are as good as GPT4, while there will still be a market for a “GPT5”, the current applications will experience serious price competition.
If OpenAI has to make a profit or at least lose less money to please investors, they will raise the price of GPT4 or flat out cancel it.
It doesn’t matter, if you can run Llama3/4 on commodity hardware is what I’m saying.
If I can fulfill my objectives with a commodity model with GPT-4 capabilities then I’m not going to pay any $$$ for GPT-5 because my needs are already met. So OpenAI will lose customers at the same time as gaining them.
The price at the top end will go up, but the price at the bottom end (which is currently the top end) will fall.
I think Meta is the last major player who's only releasing free generative AI tools, so with no incentive to kneecap their best free stuff to avoid cannibalizing their own products. AI companies will have to build actual products on top of their models to produce more value, so I want to see if Meta will release whole free products to keep the pressure on. Maybe "Segment Anything" is already a step in that direction.