News for hackers. Stuff that matters.

A comment on Meta got caught gaming AI benchmarks. Parent comment

Re:Meta got caught gaming AI benchmarks by minimaxir on Tuesday April 08, 2025 @02:08PM (#43624753)
One very important distinction between image models is the implementation: 4o is autogressive, slow, and extremely expensive.

Although the Ghibli trend is market validation, I suspect that competitors may not want to copy it just yet.

[ Parent | on HN ]
Re:Meta got caught gaming AI benchmarks by JamesBarney on Tuesday April 08, 2025 @03:58PM (#43625843)
Extremely expensive in what since? In that it costs $.03 instead of $.00003c? Yeah it's relatively far more expensive than other solutions, but from an absolute standpoint still very cheap for the vast majority of use cases. And it's a LOT better.
[ Parent | on HN ]
Re:Meta got caught gaming AI benchmarks by svachalek on Tuesday April 08, 2025 @04:14PM (#43626010)
Dall-E is already 4-8 cents per image. Afaik this is not in the API yet but I wouldn't be surprised if it's $1 or more.
[ Parent | on HN ]
Re:Meta got caught gaming AI benchmarks by echelon on Tuesday April 08, 2025 @02:16PM (#43624817)
> 4o is autogressive, slow, and extremely expensive.

If you factor in the amount of time wasted with prompting and inpainting, it's extremely well worth it.

[ Parent | on HN ]