An H100 today costs $2.95 an hour on vast.ai[1], which is already a good deal.

gpt-oss-120b on an H100 gives you ~200-250 tokens per second. I will be generous and say you can get a million tokens an hour out of it.

OpenCode Go (which I gladly pay for, because of this in part) is $10 a month, that's three hours of H100 use, and the models you have there are more expensive than gpt-oss-120b. Sure, they have "scale" (although that doesn't apply to AI inference, but whatever) and this and that, they're still pricing it 20-30x below their minimum threshold of capital expense.

Apples to apples, GLM 5.1 they sell it to you at $4.40 per million tokens, at ~50 tps in an H100 (being generous) it costs ~$16 to do a million tokens.

The math is simple and clear, they lose money.

1: https://vast.ai/pricing

I meant, buy/lease the hardware that lets you run this model, run gpt-oss-120b and measure. I did this once and it was like 10x more expensive than any hosted alternative, and $20 wouldn't get you far there.

Grab gpt-oss-120b, run it continuously and see how far 20 dollars worth of that gets you. People definitely use much more than that in a month, not just power users but regular ones, and they're using models that are more expensive to run (plus the "cloud" markup).

If you train a soldier in the US, is he unable to do those things outside the US?

Wrt. the pooler, how do you compare with pgbouncer?

I'm interested because I have a postgres instance, low-traffic but still like ... tens of r(eads)ps. I was not running anything close to the machine limits but still added pgbouncer to improve performance and didn't see a noticeable difference. I was stress-testing the machine obv., I'm not talking about the 10 rps, lol.

For context, my numbers were something like 10k rps +/- 1k vanilla postgres and like 9k rps +/- 1k with pgbouncer in front of it. So ... slightly slower but big error bars so I wouldn't say for sure. I ended up not using pgbouncer as the benefit was immaterial.

Also yeah, in case you want to check it out, it's the db that backs this project: https://httpstate.com.

And still ... there's a lot of this.

Maybe the solution to s..tposters is to do what Wikipedia does.

Some articles/topics are "protected" and new/unverified accounts cannot touch them.

>inb4 leverage

Yeah, I know leverage exists but still, you cannot go to a bank and ask them to help you acquire something 100x worth your cap.

Oh man, I'm really sorry you had to go through that :'(.

No, ChatGPT was the "oh shit" moment for me.

Anyone who had touched a computer before that knows how big of a leap that was.

This thread is meant for people who use AI, though.

I was a heavy and loyal Cubase user then switched to Ardour when it came out because of OSS.

Also Reason, you guys remember Reason? It had such a warm, characteristic sound out of the box. You could easily hear it in many songs of the time; same with Fruity Loops but for all the wrong reasons, lol.

I don't do a lot of music at the moment, but it's nice to read you here ^^.

Then I can make a meaningful comparison.

I applied for a job with them and proposed this exact thing about 8 years ago (got auto-rejected, I would've been very happy to work on it).

But I'm glad to see they finally did it.

Great post but this is what it reduces to.

99% of people on the planet work because they have to work, not because they want to work.

This is hardly news for anybody but it still has to be pointed out from time to time.