[corlinp]

This should erase any doubt that AI Labs are making $$$ on API inference.

Kimi 2.5 (which this is based on) is served at $0.44 input / $2 output by a ton of different providers on OpenRouter, 2.6 will certainly be similar.

That's about 11X less than Opus for similar smarts.

[gessha]

It’s worth noting that the US is very behind on energy infra and that might affect the cost calculations since data centers are electricity guzzlers. Also, not sure if CN has completely switched off Nvidia or still using them for training.

[Lalabadie]

Famously, OpenAI and Anthropic are devoted to increasing efficiency before scaling up resource usage.

[amazingamazing]

How does it erase any doubt? You’re implying Chinese things can’t be actually cheaper to produce than American which is laughable

[corlinp]

Most of those inference providers are American, and China is actually at a disadvantage here because of export restrictions - US companies are using newer and more efficient chips.

[amazingamazing]

If it’s newer and efficient then why is the api more expensive?

[veber-alex]

Price is set based on what people are willing to pay not based on actual costs.

[amazingamazing]

I’d believe that if they didn’t lower limits