OpenAI's o3 AI model, unveiled in December, displayed promising capabilities during its tests with ARC-AGI. However, revised estimates suggest that o3 high's computing cost per task is not $3,000 as initially believed, but potentially closer to $30,000. This change highlights the ongoing challenges associated with sophisticated AI models' operational costs. The Arc Prize Foundation's Mike Knoop noted that o1-pro serves as a better cost proxy for o3 due to the compute resources involved, hinting at potential high pricing strategies from OpenAI for future enterprise offerings.
We believe o1-pro is a closer comparison of true o3 cost [...] due to amount of test-time compute used.
This revision is notable because it illustrates just how expensive today's most sophisticated AI models may end up being for certain tasks.
A high price for o3 high wouldn't be out of the question, given the amount of computing resources the model reportedly uses.
The organization originally estimated that the best-performing configuration of o3 it tested, o3 high, cost around $3,000 to solve a single ARC-AGI problem.
Collection
[
|
...
]