KptMarchewa

I agree. They are definitely good - no issues with instruction following for example - but they miss the "intelligence" larger models have.

For implementation tasks, where I have the problem already defined and researched, or just simple task, I'd definitely use something like Luna xhigh or max. If the task is vague, or involves planning, I'd rather use Sol medium, even though it's theoretically worse on benchmarks.

It can bear the load of a few people, at least.

A lot of pessimism is caused by the relative change of inequality.

There is no realistic possibility to effectively ban LLMs worldwide, which makes advocating for this useless.

The inference providers did not raise the prices no?

Deepseek as a company can just increase prices for the crazily cheap cache they have, that's their only lever.

This is added specifically so you migrate out of those as fast as next models will be available.

You don't have to use Spring or the typical enterprise IOC that plagued 00s and 10s Java.

[flagged]

1/4 is a separate item.

>Cows are the most common animal on the farm

Chickens?

>The $480M bug that killed a trading company in 12 secs

It was a flag reuse issue. Nothing related to "performance"

This makes as much sense as Bill and William, or Bob and Robert.

Python is the language where in practice this is most untrue - maybe outside C++.

As an example, there are dicts, tuples, classes, NamedTuples, dataclasses, attr.s, Pydantic, and pretty much all of those solve similar problem (hold my data) but have slightly different properties and use cases.

Backend-generated templated web apps were the pinnacle of web development in 00s.

While I understand the idea of american private health insurance, I don't get how do you have to pay for kids. Do you have to pay for primary school too?

Three-four weeks can happen, at least here in Poland. They are pretty rare but I know bunch of people who did this, myself included. Usually for prolonged vacations in far away regions - Japan, South America, Australia etc.

It's easier to find a new job when you already have one.

It's not in-between, it's unemployment.

What the OP likely meant, is leaving a gap between leaving a job and starting next one. It's usually easier to push employment date by a month than getting a month-long vacation after starting the job.

At my company the performance improvements channels has been exploding, with people claiming giant improvements in latency, throughput, and decreased cost of the services. The cost decreases itself is order of magnitude more (at annualized run rate) more than we pay for tokens. YMMV.

Existing tests.

In this particular case tests won't help you, without the actual data. But overall you're right.

Exactly. In the above case, AI should be able to verify it's own hypothesis by having access to the database's table statistics and planner.

"find every function in this codebase" is rarely a thing imo. Why would you want to do that, for some statistics? In many cases, what you want to do is find "all classes that implement some interface" - of course depending on the language. And reasonable languages allow agents to find that reasonably efficiently ("implements Foo") - a bit slower than specialized interfaces, but in a way that don't require you to maintain a giant index.

And yes, agents writing chunks of code or scripts to accomplish tasks is already what they do, when they determine it's efficient.

>And there is a lot missing in the actual harnesses frankly that isn't about more parallel agent ninjutsu. Like why isn't there a 'move this function from this file to this file' tool (copy/paste char range) and we have to see Claude/Codex/etc flail around rewriting huge chunks of code in token space

I think if we had agent-specialized languages that you would modify by directly changing AST or some other structure it would be more efficient. But, otherwise, it's way more efficient to train on existing corpus - like PRs - which is the same reason that internal LLM thinking is not binharic, but regular English (or Chinese).

Do you really think it's either or? I certainly don't use my days off for that.

Well, everyone saw the freeloader class in Catholic countries too - clergy.

Yet they bend under fascist requests like current US administration.

Part of Claude is running on Elon's GPUs.

We know it's mechahitler responding.

I have not for couple of years, but before covid I've worked in a well-known Polish company that was trying to force everyone to use at first HipChat, then some awful OSS chat. My team set up discord then.

Can you take a look at actual satellite map and tell me where all the unused land you apparently see is?

Pretty much all of the land in Poland is used - either for agriculture, forests or human settlements.