I tried to build a deck with my smartphone, it couldn’t drive a single nail.

It doesn’t take much explosive to defeat infantry.

Sure, it isn’t flying across the country to hit an installation like a Shahed, but it could function as a longer-range, loitering mortar round.

Lots of small payloads would be useful if you can put a lot of them in the right place: blast fragmentation, HEDP, thermobaric, smoke, illumination, etc.

This guy has the right idea:

https://en.wikipedia.org/wiki/ElonJet

Though his posts on Twitter have a 24h delay due to their change in community guidelines.

You can get live information from their website: https://grndcntrl.net/

Companies just have way too much data about people.

It should be illegal to store/compile data that isn’t directly related to the good, services or products that you’re offering.

We’re getting to the point where everything about you, down to your real-time location, is available to anyone with enough money. That’s just not a power that we should allow to go unregulated.

Yeah, there are lots of ways for this to be true but misleading:

The communications are not encrypted if they have the keys.

The encrypted communications are not the people’s. By the TOS everything is the property of WhatsApp and they can access their own ‘Business Records’ perfectly legally.

A third party, like a federal agency, isn’t WhatsApp. (WhatsApp can also voluntarily give their ‘Business Records’ to said agencies without warrant or subpoena.)

Meta isn’t WhatsApp.

An internal project with an undisclosed codename isn’t WhatsApp.

You can’t compare the cost of running a model at home to the cost of running a model operationally as a business.

I didn’t say running it at home, that’s ridiculous. The Pro model would not run on home hardware.

I said:

compare that to the cost of running the DeepSeek model on comparable hardware

Here, comparable hardware means an NVIDIA H100. A card who’s use has a well-known market price.

Why the If, we know they are running at huge losses.

Ok and I addressed that also:

if they are eating a loss then that means their inferencing is even more expensive so the DeepSeek model is even better than 1/6th the cost.

They are selling their inferencing at a loss, and therefore their inferencing cost is higher than the amount they charge. The 1/6th number comes from comparing the amount OpenAI charges to the amount it costs to run DeepSeek.

Since their costs are higher, then the ratio of their price to DeepSeeks is even better than the quoted 1/6th figure.

Because, in mathematics, if the numerator is fixed (the top number, i.e. the cost to run DeepSeek, which is known because you can run it yourself on the exact same hardware) and you increase the denominator (the bottom number, representing the cost of GPT/Claude) then the ratio becomes smaller.

Since you agree that they’re losing money on inferencing then that means the bottom number is unknown, but we know it is higher than the price listed on their website. So, the 1/6th ratio represents the upper bounds on the ratio of costs.

That doesn’t matter.

We know DeepSeek costs 1/6th as much as the cost that OpenAI/Anthropic are selling their model’s tokens. We know this because we can see their prices online and compare that to the cost of running the DeepSeek model on comparable hardware (because it is open weight).

If OpenAI/Anthropic are selling at a loss then it means their internal inferencing costs are even higher than the price that they list on their website.

This means that the DeepSeek model is at least 1/6th the cost, assuming that OpenAI/Anthropic are selling at cost. If they are eating a loss then that means their inferencing is even more expensive so the DeepSeek model is even better than 1/6th the cost.