cbg0

Is the prompt generated by the LLM to send to a sub-agent really your data?

Isn't AOSP a thing?

In the past they received a lot of hate for not comparing to the competition.

5.6 Terra (mid tier model) as good as Fable on DeepSWE while cheaper than Opus API pricing. Seems like a homerun.

Hy3 DeepSWE - 28%

GPT5.4 xhigh DeepSWE - 52%

A lot of contaminated benchmarks in the blog post about Hy3, needs real testing though I have a distinct feeling it's benchmaxxed like a lot of Chinese models.

You should be relieved that they're sending robots instead of you to get blown up by a drone.

Assuming these concerns were real you could just turn the app into a website.

It looks like some of these systems do work, why shouldn't we try these systems too?

https://www.sciencedirect.com/science/article/pii/S002243752...

> How would these laws prevent me from just side loading my own open source client?

They do not.

This will not get triggered by blind spot checks and most new cars have these inside the mirrors, so you won't even need to look over your shoulder.

Personal freedom to fall asleep at the wheel or look at your phone while driving?

It works, but there are mechanisms to also track your head position if it can't see your eyes.

The GDPR protects people from companies keeping their data and using it recklessly.

Perfect is the enemy of good. If it prevents some people from falling a sleep and causing an accident that's excellent.

Preventing car crashes and people dying in said crashes isn't an unhealthy obsession with safety.

Better in what way? Does it follow the goals better, does the code produce have higher quality in a testable/maintainable sense or is it just closer to how you would usually program something?

I've seen this with GPT, and I usually ask it to put together a more easy to understand document for a specific target audience or reading level and it seems to do okay.

> This is because GPT-5.6 is just a more posttrained version of GPT-5.5, not a bigger model than GPT-5.5.

What is this very confident assumption based on?

I'm neither an old man, nor am I grumpy at the new tools. I use AI daily, I'm just not deluded into thinking AI is more capable than it is, or myself for that matter.

> We're accelerating at lightning speed now.

Accelerating how much slop you can output? A better model will still produce slop for your feature factory that pumps out software which nobody is interested in buying.

It's a post on twitter, not a press release.

Talk about overreacting.

This is an intermittent issue, you should still be able to get your work done. 5.5 was released two months ago so perhaps you're using 5.5 wrong and some things that worked in 5.4 require tweaking your prompts?

Doesn't look like it: https://marginlab.ai/trackers/codex/

You want a refund because of a problem you weren't even aware of until now? And you don't even really know if your work has been impacted by this problem.

I'm not sure I follow; they were given 90 days to comply with certain requirements from the EU, which they did, and the fine was something extra that took a while to resolve in the courts.

This approach is more US-centric than "HN people". I've heard this often with phrases like "better to ask for forgiveness than for permission" which is probably because you can get away with a lot more on the business side in the US without having personal accountability.

> No, they're OK with Google apparently not changing much of anything and being allowed to continue operating so long as they pay the fine first.

This is false. They were asked to:

- Stop tying Google Search and Chrome to the Play Store.

- Permit competing Android versions.

- Stop exclusivity incentives for Google Search.

- Provide genuine room for rival search engines and browsers.

This is separate from the fine and they were given 90 days in 2018 to comply with the above.

The point of fines is to correct behavior, not to cause financial ruin.

Do you really believe the EU needs to "disguise" taxes? And that a $23 trillion economy wants to wait 8 years to collect $4.7 billion?