I am suggesting a change from the o3-mini models to o4-mini models.

:

MODEL COMPARISON:

o4-mini o3-mini

----------------------------|-----------------------------

Knowledge: 2025-04-16 | Knowledge: 2025-01-31

Input: $1.10 | Input: $1.10

Cached: $0.275 | Cached: $0.55

Output: $4.40 | Output: $4.40

The cost would be a wash for input/output, however it is 50% less for the o4-mini for cached input.

Now I don't know how the backend is structured, however I am reasonably sure cached inputs are being utilized, which could be a nice cost savings with a rather sizable improvement for the customer base.