The players barely ever change. People don't have problems following sports, you shouldn't struggle so much with this once you accept top spot changes.

It is funny seeing people ping pong between Anthropic and ChatGPT, with similar rhetoric in both directions.

At this point I would just pick the one who's "ethics" and user experience you prefer. The difference in performance between these releases has had no impact on the meaningful work one can do with them, unless perhaps they are on the fringes in some domain.

Personally I am trying out the open models cloud hosted, since I am not interested in being rug pulled by the big two providers. They have come a long way, and for all the work I actually trust to an LLM they seem to be sufficient.

I find ChatGPT annoying mostly

Open settings > personalization. Set it to efficient base style. Turn off enthusiasm and warmth. You’re welcome

Yea but even then it's still annoying. "It's not about the enthusiasm and warmth but the general tone"

Setting “base style and tone” to “efficient” works fine for me.

Their financial projections that to a big part their valuation and investor story is built on involves actually making money, and lots of money, at some point. That money has to come from somewhere.

I didn't express this well but my interest isn't "who is in the top spot", and is more _why and _how various labs get the results they do. This is also magnified by the fact that I'm not only interested in hosted providers of inference but local models as well. What's your take on the best model to run for coding on 24GB of VRAM locally after the last few weeks of releases? Which harness do you prefer? What quants do you think are best? To use your sports metaphor it's more than following the national leagues but also following college and even high school leagues as well. And the real interest isn't even who's doing well but WHY, at each level.

The technical report discussing the why and how is here: https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro/blob/main...

Follow the AI newsletters. They bundle the news along with their Op-Ed and summarize it better.

Can you suggest some good ones?

Thanks for this!

Link to direct newsletter subscription: https://importai.substack.com/

I really like latent.space and simonwillison.com.

Also (shameless self-promo) I publish a 2x weekly blog just to force myself to keep up: https://aimlbling-about.ninerealmlabs.com/treadmill/

Tips on what newsletters are worth signing up for ?

holy shit im right there with you

Don't keep up. Much like with news, you'll know when you need to know, because someone else will tell you first.

This is only good advice if you don’t have the need to understand what’s happening on the edge of the frontier. If you do, then you’ll lose on compounding the knowledge from staying engaged with the major developments.

Not all developments are equal. Many are experimental branches of testing things out that usually get merged back into the core, so to speak. For example, I knew someone who was full into building their own harness and implementing the Ralph loop and various other things, spending a lot of time on it and now, guess what? All of that is in Claude Code or another harness and I didn't have to spend any amount of time on it because ultimately they're implementation details.

It's like ricing your Linux distro, sure it's fun to spend that time but don't make the mistake of thinking it's productive, it's just another form of procrastination (or perhaps a hobby to put it more charitably).

I agree that a full linux distro compile as a matter of practice is a waste of time. But, doing it a few times is good if you want to understand your tools.

I don’t believe that top tier engineers just skip learning things because they might turn out to be dead-ends or incorporated into tools by someone else; in my experience they tend to be extremely interested in things that seem like minutiae to others when working on the bleeding edge, often implementing their own systems just to more fully understand the problem space.

If it’s a day job for someone and they are not ambitious, fine. But we are at hacker news. I would bet 99%+ of top tier software talent could tell you practical experience with ralph loops this year, or a homegrown variety, simply because they are an attempt to solve a very real engineering problem (early exit, shitty code/incorrect responses, poor context window length and capacity), and top tier software people expect more control of their engineering environment, and success using their tools than they’d get by just saying ‘meh, whatever, I don’t get this and I’ll just wait it out.’

This one’s been particularly hard to sit out because the executive and managerial class are absolutely mainlining this stuff and pushing it hard on the rest of the organization, and so whether or not I want to keep up, I need to, because my job is to actually make stuff work and this stuff is a borderline existential risk to the quality of the systems I’m responsible for and rely on.

Thus, in the situation you described, "someone else will tell you first" is your boss.

It honestly has all kinda felt like more of the same ever since maybe GPT4?

New model comes out, has some nice benchmarks, but the subjective experience of actually using it stays the same. Nothing's really blown my mind since.

Feels like the field has stagnated to a point where only the enthusiasts care.

For coding Opus 4.5 in q3 2025 was still the best model I've used.

Since then it's just been a cycle of the old model being progressively lobotomised and a "new" one coming out that if you're lucky might be as good as the OG Opus 4.5 for a couple of weeks.

Subjective but as far as I can tell no progress in almost a year, which is a lifetime in 2022-25 LLM timelines

Another annoyance (for more API use) is summarized/hidden reasoning traces. It makes prompt debugging and optimization much harder, since you literally don't have much visibility into the real thinking process.

Opus 4.5 was released on Nov 24 last year. It’s only been 5 months!

Wow you're right, okay not so bad then.

That brief two week period when Opus could eat entire tickets was simultaneously fantastic and a bit alarming

I don't trust the benchmarks either, so I maintained a set of benchmarks myself. I'm mostly interested in local models, and for the past 2 years they have steadily gotten better.

Can't argue with subjective experience, but if there were some tasks that you thought LLMs can't do two years ago, maybe try again today. You might be surprised.

I’m very satisfied with being three months behind everything in AI. That’s a level that’s useful, the overhyped nonsense gets found out before I need to care, and it’s easy enough to keep up with.