And serving is not training. For distilling you need to train the big models to have something to be distilled.

But that i pay for some 2tb storage and i'm a 'pro' user while not really a 'pro'user and that there is another 'pro' package makes all of that very weird. This is something they need to clean up

They use Gemini for personal assistent to all of their Gmail and co users/customers. They have Google Docs, they have GCP were gemini should support you too.

They also have a lot more languages to support too.

They optimize Gemini for A LOT more than 'just' coding. So its probably a balance act for them. And because they are that rich and have no issues on compute and brain, they can play the long game easily.

If they push their tpu further and continue their build out, they will be able to start training high quality topic optimized models in parallel while everyone else needs the same amount to just train one main model.

Nonethless, in comparision to a lot of other companies, Google has Deepmind and the money to just do all of it without spending money which doesn't exist.

This alone is such a crazy time to experience the amount of compute our society is adding globally.

Geminis quality increased again + long running agents. This will hurt so many people who do 'stuff on a computer'.

If this hits production, i probably would prefer to fire 1-2 collegues and get their salary in tokens. And its only a question of time until someone has the same thoughts about me.

Nothing of all of that is underwhelming.

How can we have such extrem opposit viewpoints to this?

But at least for me, I really read a lot of code, a lot of PRs from others and evaluated them.

You could literaly just give your compute away for free for a year to pull people in.

Make an API Endpoint for free with the caviat that they are allowed to use the data for traing, what everyone else does too.

I'm a Cloud ARchitect with experience in coding (15 years) and infrastructure (10 years) and startup founder...

If you don't comprehend what i write, feel free to ask but don't be dick?

2 years ago when I prompted something, it had compile time errors left and right. Took me 3-10 iterations to even get it running.

Now its one shoting a lot. Including websides, refactorings, etc.

The question is what is missing? How far are we that it can handle huge code bases vs. smaller ones? How far are we that it can comprehend the whole architecture and doesn't try to put a service in a wrong place just becaus the context is too small?

Mythos is 10 Trillion, that might be already pushing it.

95% might be not enough for someone in sense of "yeah i can't do the 95% and i can't do the 5% either the AI can do 100% or i still need Kevin with his knowledge even if its just for the last 5%"

Your Product Manager is not a coding job. Your Product Owner is not a coding job.

vibe-kanban exists you could already do a proper experiment letting your PO maintain a vibe-kanban board with proper requirements and see how an agent progresses.

But 5% is often enough wwhat breaks it. Doesn't help much when your PM, PO or CEO or CTO have no clue about coding harnesses, coding agents, coding platforms, LLMs etc.

For this amount of money you can rebuild cursor and everything else on the market, and with the rest of 9-59 Billion, you just hire experts in coding and let them code real high quality code examples.

And then you just use your existing grok pipeline and just add this functionality.

This xAI stuff has to be run by idiots

CRUD applications and converting business requirements into code is the thing software developers do to 99% day in day out.

Dude what the hell happened to Musks Grok? How incapable are they that they give away training compute to Cursor like this?

Weird that the genius Musk doesn't need his own compute, after all shouldn't Macrohard (no joke) already building the worlds software from scratch?

Its 'production' code because its a small browser game which has very small to 0 requirements on security and being perfect but high requirements on 'ever even doing this' and 'fun'.

The code it generated hat 0 compiletime errors. I was able to descripe 10 things to do in one task and it just jugged along solving all of them.

This doesn't need to become so much better to be useful. Its already very useful for a lot ofuse cases like researchers which have to verify the math anyway but are not good in writing code for filtering their testdata, converting them and running it.

Small websites, fun projects, helper tools etc.

But while we speak, in the background stuff is still happening left and right. More compute, better algorithm, more RL etc.

We could already be at 95% at 'ai will take your coding job' without knowing because these 5% are so relevant.

The good thing though: the effort is low. You think through it once and you have your encryption and backup strategy for a long time.

I have a NAS System which only runs when i need it, i scrub every month and that basic setup is the same for the last 12 years.

Every machine is encrypted, unlocked per login.

Encryption is basically free so.

I do not want someone stealing my laptop on a train ride potentially being able to have all of that data.

With a proper real backup strategy, i have everything save. I do not need easy access to a hard drive from a broken computer.

But hey you do you :)

Like how about you have to do a short course which actually explains to you how a drug works, how to use it correctly, what are potential downsides, what are markers of overuse/wrong use.

And the other main issue with opioids and co: some people really have constant / chronic pain.

Do you know how exhaustive it is to constantly have pain? How annoying it is that you can't just go to bed and sleep?

But also we can't play devils advocate to say "you are not allowed to do drugs to num whatever issue you have" and also "but i don't want to take time and effort of helping you".

Oh i don't want you to kill yourself! But i don't want to spend time tomorrow afternoon either with you.

Our society is very hypcritical in this sense. Honestly i think people just don't want to see homeless people or fentapoeple. Its not about helping, its just about not being disturbed by them.