prasoon2211

I used to be on 5.4 high for most of my work. I have switched completely to 5.5 medium now. I would highly recommend trying it out

- 5.5 is significantly more token efficient than 5.4 - the same task takes often a third of the tokens

- because of this, is it also much faster to do the task

- you get high "intelligence" per token even after accounting for token efficiency - 5.5 medium is just under 5.4 pro levels of intelligence (imo). It has found tricky bugs for me that all other models failed at

So overall, ideally you will end up with more intelligent, faster model for slightly cheaper.

May I ask if you're an insider / some who has first hand information about this?

I founded a UG and a GmbH in 2024. It took me 3 months total including visits to the notary (who charges a non-insignificant sum for their services).

I did this as a subsidiary for a US company and literally had to email and call people every few days to move the process along (mostly, it was the banks who somehow expected us to be a multi-national company and wanted to charge an arm and a leg just to let us open a bank account. Most banks outright refused us).

When the notary finally filed the paperwork to the court, the court replied after a few weeks with additional clarifications for which we had to go AGAIN to the notary to do the whole song and dance of them chanting at us in German at 1000 words per minute.

Everything took painfully long and delayed investment for while. People have absolutely no idea how painful it is to merely have the incorporated entity available. Then, it takes a few weeks to get your tax ID - this is when you can start employing people / accepting payments etc.

This is incorrect. Taxes, pensions and labour laws apply according to the country where the EU Inc is operating.

iirc he (or his colleague) did mention somewhere on X that most of the PRs are small

Agree on all points, especially half of my friend circle being unemployed at one point.

Folks with amazing jobs having to spend >1y trying to find a "meh" job. Pretty much impossible to find a job without fluent German.

Most people actually using python do not start off in scripts. Usually, I would mess around in IPython / Jupyter for a couple days until I have something I'm happy with. Then I'll "productionize" the project.

tbh this has been a sticking point for me too with uv (though I use it for everything now). I just want to start of a repl with a bunch of stuff installed so I can try out a bunch of stuff. My solution now it to have a ~/tmp dir where I can mess around with all kinds of stuff (not just python) and there I have a uv virtualenv installed with all kinds of packages pre-installed.

This is partially why, at least for LLM-assisted coding workloads, orgs are going with the $200 / mo Claude Code plans and similar.

Presumably the tab based edit-prediction model + $5 of tokens is worth the (new) $10 / mo price.

Though from everything I've read online, Zed's edit prediction model is far, _far_ behind that of Cursor.

RLHF is not the "RL" the parent is posting about. RLHF is specifically human driven reward (subjective, doesn't scale, doesn't improve the model "intelligence", just tweaks behavior) - which is why the labs have started calling it post-training, not RLHF, anymore.

True RL is where you set up an environment where an agent can "discover" solutions to problems by iterating against some kind of verifiable reward AND the entire space of outcomes is theoretically largely explorable by the agent. Maths and Coding are have proven amenable to this type of RL so far.

Right, it took me a couple of re-reads and I (non-native speaker) ended up asking ChatGPT about it and yes, the sentence is worded incorrectly

As others have said, this is classic selection-effect. Lina Khan isn't coming out and telling people about

- the companies that died because acquiring them was too much of a hassle

- the companies that died or never got funded / started because the investors couldn't see and exit path

- the companies that got acquired piecemeal (Windsurf, Inflection), leaving the early employees with NOTHING simply to avoid the ire of anti-trust hawks at the FTC. This has irreversibly damaged the SV bargain - early startup employees work hard in case of an acquisition, they get rich.

So Lina Khan can keep patting her own back but there's a reason founders, early-stage startup employees and investors disagree.

I worked at a company where we copied this from Amazon for a specific type of meeting (bi-weekly review). But we also had the other "normal" type of meeting.

People never read the documents before the meeting in those "normal" meetings.

The challenge with your suggestion is that people will half-ass the doc reading before the meeting - we tried doing this for the "normal" meetings. It was obvious the people skimmed the doc before the meeting. You're also now relying on the manager (if there even IS one for everyone in the meeting!) to care about this.

So, in practice, giving people dedicated 10 minutes at the start of the meeting works far better.

Besides, in most "normal" meetings, the main presenter often ends up discussing background / context for 10 minutes interspersed throughout the meeting anyway. In the "pre-read" meetings, you're just compress that to the first 10 minutes while increasing the amount of information transferred.

This is really cool! Just started using it today. It's missing some of superwhisper's ease of use but other than that, 10/10

I tried using it for something non-trivial. And:

> 429: Too many requests

Mind you, this is with a paid API key