ben8bit

Terra is also a fantastic model.

I really liked their brand position a year ago. But having Boris talk about "Claude's brain" on stage & their other founder pretending AI is sentient in front of the pope - they're coming off more as a religion than a tech company (and this is aside from the obvious model issues they're having).

So what we have here is the LLM version of the aesthetic usability effect. So: prettier = better.

Take from that what you will - and maybe the AUE is more desirable - but Sol & K3 are genuine game changers for existing codebases. Anthropic currently have an antagonism problem which has worked for them in the past, but not anymore I think, as other models have become as competent.

Sol is absolutely incredible. It's the first model where it feels like a mid-weight engineer that actually looks at the details. I haven't tried Max yet - High for me has been enough so far.

Grok 4.5 is actually good though (as a workhorse model).

Interesting, I generally use High for all models, even non Anthropic ones.

This was one of the few times I tried Max, and the problem was the code (it listed as a "hole") was directly adjacent to the problem area, and not especially complex.

Kind of like looking at a washing machine and telling the customer to be careful because the inlet pipe will pump water into an empty box.

6 months ago: get Claude to do some work, have GPT review it, ask Claude to first verify findings before actioning.

Now: get GPT to do some work, have Claude review it, question Claude about a finding that is surprising to me because I thought the functionality was already in place.

[Claude/Opus 5 Max goes looking] "You're right — I was wrong about that."

Our Claude license ends in about 2 weeks and we're not renewing - this has been par for the course for the last 2 months now.

And their marketing is really starting to bother me, on top of that.

Really? I've heard so many other people complain about this recently. And maybe it's possible that it's the prompt style even. But interesting that it's not across the board.

Haven't tried this yet, but going to soon! I have to wonder what happened at Anthropic. We've cancelled our subscription in favor of OpenCode & Codex. Sol is just so good & OC goes so far for every $ spent. Claude's become a pain to work with - average output with an annoying personality. Who knew this would be an issue even a year ago? In any case, loving the stuff from the Chinese models!

Honestly, I didn't even know this was a problem that needed to be solved.

More and more these companies sound like they want to protect market share, instead of lose it to foreign models. I believe the cat's already out of the bag on this one. Regulatory capture and more export control will indeed push more people to Chinese models that are good enough right now. I'm not sure they actually realize that.

I'm surprised by the negative its-a-copy sentiment.

Of course it is. Just like Linear/Mailchimp/etc. (almost everything) are copies of other things. These harness+work apps are becoming a product category and I'm excited about this from the Kimi/Moonshot team. In the end, more competition in this space can only be good for the user.

Probably a good example why you shouldn't outsource creative work to AI.

I mean, it's hard not to be impressed by the Moonshot team. Absolutely great work.

I'm surprised - weren't they always OSS?

Definitely a +1 from me. I've really enjoyed using it via OpenCode/Zen. Not loving the pricing with OC so will probably switch to OpenRouter once my credits are done.

Brave.

The biggest problem we face in the west is thinking our institutions are somehow different. Be critical of the product all you want, but don't pretend the exact same thing isn't happening here.

The tooling has become so good though - the eco-system around the LLM. The models have become really good, yes - but it's definitely slowed in my opinion. The tooling is what really has become great - "harness" is probably the best word. When folk like Elon/Schmidt/Theil/etc. talk about singularities and industrial revolutions - it sounds extremely out of touch - or actually protective of the massive capex they've potentially sunk.

EDIT: Schmidt's booed commencement speech was probably one of the most out-of-touch speeches (outside of a tech interview) I've heard.

The entire problem with "AI" is that it's easy to do without. The AI companies know it, the users know it - even the most pro AI agent manager knows it. Thought experiment: remove AI from the world right now, all of it - what do you have? Business as usual. This article doesn't do enough to underscore that - dreaded be the day I need to get an actual engineer to review a PR, right?

Why is this phrased so antagonistically? For clicks? This is like saying the US is winning the cloud race - it's not a race. The author is completely missing the forest from the trees - this benefits everybody. LLM's aren't going to wake up and be sentient - the race to real AGI is where is has been 10 years ago.

It's pretty interesting because if you look at the net value over the last 2 to 3 years, you'd expect to see a flurry of high value/high complexity/high velocity software being delivered. But we haven't. We've seen outages almost normalized now and the only new thing being built are more AI integrations - and I have to ask: for whom? To me, there's a pretty big gap between the delivery and claim of AI, and the question is - do you stay an IC or do you become an agent manager (in which case you will lose your technical edge for sure).

I would not be surprised if AI commits are the culprit. There is no way any service would cope with a constant stream of unfettered commits by sleepless always-on agents. Ironically, this same strategy seems to be what GH/MS (and other big companies) are evangelizing - and therefore dying by their own hand (in a way).

I think picking up people at the bar is easier than making friends at the gym - what you want is to join a crossfit gym, or something that has a stronger community culture to it. Not the gym.

I love coffee, so this is a nice read. Couple years ago I switched to french press, fresh beans (grind on demand) & no milk or sugar - okay, a dash of full cream milk sometimes. Has to be strong - you can't drink weak coffee like that!

We initially looked into using shadcn/radix - but it's quite bloated. Maybe even warranted with all the a11y accommodations - even the official W3C ARIA examples are not straightforward (most of the time).

Yea, was wondering this too.

> Apple hardware isn't exceptional, it's maybe slightly above average

Is this a serious statement? If Apple's hardware is maybe slightly above average - what's above it? It's an easy company to hate on - but you don't have any other platform integrated as well as Apple's right now IMO. Unless you maybe count Huawei.

Edit: I think I may be referring more to the holistic picture. But still curious what hardware you think is better.

For all of the Apple hate recently (with glass + MacOS bugs) - it's still great to be able to invest in high quality hardware. I made the switch to Mac, oh about 12 yrs ago, and sometimes forget how spoilt we are with really tight hardware + software integration.

Yea, it's a very odd way of referring to people you work with.