4 ms·
Luckily inference is cheap, and other providers offer efficient models. It’s only going to get better in the future.
by user34283 5mo ago
Luckily inference is cheap, and other providers offer efficient models.
It’s only going to get better in the future.
- ai_fry_ur_brain 5mo agoIts not though, just because your favorite CEO or youtuber said it will, doesnt mean it will. Inference is not cheap, you have no idea what you're talking about. Every new chinese model has doubled their prices in the last two weeks
- user34283 5mo agoKimi K2.6, the leading open-weights model, is $0.95 input / $4.00 output / $0.16 cache. That’s about 1.45x more expensive than K2.5 from January. It is around 5x cheaper than GPT 5.5.