Oh this has already been clear to anyone in the EU, for example. The current reliance on US tech and even widespread stuff like MS is pretty deeply rooted, however and it might take a while to do anything about it - so for many it’s a matter of convenience for now.

That said, as long as what you need sits behind an OpenAI or Anthropic API and you don’t have deeper proprietary integrations, there is no moat. I can even run Claude Code with DeepSeek if I so choose (though OpenCode is neat too).

Best EU has at the moment seems to be Mistral though, which is… sorta passable, but not cutting edge. Oh well.

> I think the right move for Europe and other countries would be to effectively ban US tech and follow the Chinese response to Nvidia (delivered personally to Trump: we want to build our own AI chips).

Not sure about outright ban, but homegrown govt. systems should have both the devs and the infra in EU.

Would also be really cool if we could make even regular CPUs and GPUs some day but I don’t think that’s super likely, though. Kinda amazing that China can do that! Even consumer stuff like the Chinese Lisuan GPUs (and Moore Threads I think), hell, even the Russian Elbrus CPUs.

Anyways, this seems like pretty good PR for Anthropic: “Our models are so powerful even the government forbid us from exporting access to them as a service for a while!” for once this gets sorted out (if it does). It’s one thing when they just write self-congratulatory blog posts and people are skeptical, it’s another (at least, optics wise) when the government targets them, specifically.

Ofc the original intent might have been to hurt them by removing their advantage vs OpenAI, go figure. I wonder whether OpenAI's next models would get a similar treatment, or whether the govt. would also decide that Opus 4.X and GPT-5.5 shouldn't be given to foreigners as well. Who knows if some money needs to change hands behind the scenes in the form of a charitable donation.

If this affects all LLMs long term though, things will be pretty messed up.

Yes, and the lack of a Recycle Bin of any sort is even more puzzling. I think both servers and desktop PCs across all OSes should have it by default, so unsafe deletes would be something you'd have to go out of your way to even enable.

“What AI agent?”

The science fiction writes itself.

Unfortunately sometimes my Intel Arc B580 has a driver quirk where all the windows freeze and unlike Chromium based browsers I can't open Task Manager and kill the GPU Process and have it restart and have everything keep working, but rather have to kill the whole browser and restart it and hope the tabs load back correctly - thankfully haven't had many issues with losing those (only once or twice in the past year, but those were fucking annoying).

Either way, I explored both Edge as my daily driver for a year or so and also Safari on my Mac - both are actually fine as far as the user experience is concerned, but in the end I still come back to Firefox. It's a browser, it doesn't feel user hostile, it does its job well enough. Also personally I like its devtools more.

I think they might be hitting a point where subsidizing the expensive models for subscriptions makes less and less sense.

With Opus 4.X, last month I paid 100 USD for the Max subscription and got a token equivalent of 4.1k USD.

I imagine that Fable is more expensive to run.

What if for a lot of software projects out there you don't need craftsmanship but you need something closer to real engineering: "This is what a sane PostgreSQL setup and transaction management for your app looks like, this is how you do validations and the ORM layer and logging and interaction with APIs and request queueing, this is how a good front-end page looks like, here's the off-the-shelf component library you use, here's how your process errors and show toast messages and handle redirects without messing around with browser history, here's all the accessibility and mean things you DON'T do (like hijacking scroll, not having contrast, having too many animations etc.)."

You don't have a rockstar building a bridge, nor do you usually have craftsmen taking risks on innovative materials and new approaches: most of the time, you just build the damn bridge in ways that have worked for decades and have been proven. Software industry doesn't seem to know what is proven or works, cause it's a moving target. Outside of maybe niches like writing code for airplanes, completely different standards there, not sure if most devs would personally want to work in those conditions, though.

Instead on one hand you have orgs and devs that are moving too fast and create a lot of churn without nailing down what works and doesn't, on the other hand you have a lot of people (myself included) that often have to push out slop because tech stacks are a mess and you still have deadlines which don't care about any of that, and largely the state of software development seems like a comparison between Windows 9X, the UI/UX of which was at least partially based on usability studies, and the modern version which... well, you can see for yourself.

At least with AI that actually does things more quickly, there is a bit more breathing room (introducing AI is easier than changing a given environment).

Aside from that, I wonder how much variety there is in practice: between "Oh yeah, I added that new button while we were in the meeting" and "The new button feature will be ready in Q3 according to the roadmap, once we have sign-off from all the stakeholders."

It’s also quite affordable, at my current usage the DeepSeek tokens cost approx. the same as my Anthropic Max 100 USD subscription, though that’s also because DeepSeek generally needs more tokens.

I’d say I have fairly moderate usage, the DeepSeek dashboard shows around 100 million tokens per day, but almost all of it cache. Without cache it’d be like 1.5 million in and 0.5 million out most days, sometimes double, other times half.

Used it with Claude Code for a while, though I have to admit that using OpenCode with DeepSeek just sparks joy. Tone wise, it’s also a bit less obnoxious than Opus sometimes, though the flip side is that it’s wrong more often and sometimes just does dumb shit when it comes to code.

Meanwhile, some projects are doing the opposite, like going from Rust to Zig, here's an example from a podcast I recently listened to: https://www.youtube.com/watch?v=XSXGf3oN2yU

Here's the project in question: https://github.com/roc-lang/roc

I think Bun just got a lot of visibility because of the speed and scope of the migration, which both shook things up and I guess was good PR cause that made a lot of headlines.

> v3.4.3 has been out long enough that its rate (5.00) is already comparable to historical releases. The "wait and see" argument is an appeal to an unknowable future that shifts the burden of proof away from the critics. If more bugs surface, they will enter the distribution like every other release. There is no reason to expect a regime break.

I mean, as someone who uses LLMs, it might be a good idea to consider how one might limit the amount of bugs that will appear in the future at least a little bit: parallel iterative code review loops would probably be the easiest and most applicable to LLMs, though I guess test coverage and other code analysis tools help too.

1) Claude Desktop which includes Claude Code for Anthropic: https://claude.com/product/claude-code (alternatively the terminal based version; either way get the subscription)

2) Codex for OpenAI: https://developers.openai.com/codex/app (same as above, subscription preferred instead of paying per token)

3) OpenCode for a variety of models: https://opencode.ai/ (they also have a subscription, but this in particular also makes it really easy to connect to OpenRouter)

4) KiloCode is essentially the above, but for VSC derived editors: https://kilo.ai/ (I personally liked RooCode more, but that got retired)

More niche tooling options:

1) Zed is pretty good, though I saw some issues with their LSP Edits and found that connecting them to OpenCode through ACP worked better, still a cool editor: https://zed.dev/

2) If you have to pay for tokens and can't get subscriptions, look at DeepSeek as a provider (V4 Pro with Max reasoning): https://api-docs.deepseek.com/quick_start/pricing

3) I'm also writing a launcher to make running Claude Code with 3rd party providers earlier, early days still: https://ccode.kronis.dev/

Note: for anyone on Windows, if you install the terminal versions of the tools (Claude Code, Codex, OpenCode, ...), you probably want them inside of WSL so there's less confusion with file paths etc. that some models have.

In regards to actually using the tech:

  - version control and maybe worktrees
  - sub-agents are pretty nice to have, Claude Code also introduced support for longer running workflows
  - throw as much tooling as possible at the project, like Oxlint, Oxfmt etc., for Python it might be Ruff and ty or Pyright or whatever
  - throw as much testing as possible at the project, maybe require certain coverage or just have CLAUDE.md that nudges the models to write and run tests
  - throw as many additional scripts at the project as you want, e.g. how you want the architecture to be laid out, max file length limits etc., whatever common tools don't cover
  - some tools also support LSP, use those when possible
  - pretty much all models will still output slop, though making fresh instances (even of the same model) review its output, e.g. 3 parallel sub-agents looking for critical/serious issues works pretty well, I just have a review loop that I make the models run before commits
  - ideally you'd also test local instances of whatever you build (e.g. real PostgreSQL instance etc.), just so the dev loops are tighter and faster

Even here, companies don't really trust Eastern providers that much, so they'd be looking for someone in the EU running DeepSeek instances, which might come with a bit of markup. Those orgs would also sometimes be weary of OpenRouter which to me seems like shooting yourself in the foot by being so picky.

That said, DeepSeek V4 Pro (with Max reasoning) is pretty okay and I'm using it instead of Opus 4.8 (my Max 100 USD subscription weekly limits ran out today) and it can do stuff passably (even better than Mistral's offering and has nice context window), but compared to the amount of work I can get done with Anthropic's models, it keeps occasionally fucking up and I have to go back and correct it, so lots of token waste. Maybe it's close to SOTA from 6-12 months ago, though, which is pretty cool on its own, though - just less confidence in its output.

It's like trying to limit the costs and therefore not gaining the maximum added value from the technology. Similarly for those trying to run stuff on-prem, we don't really have the electrical grid here for large scale inference in-country, nor is anyone exactly salivating at the idea of dropping multiple tens of thousands of EUR to build out something passable. I do host some stuff on a bunch of Nvidia L4 cards (Qwen3.6 35B A3B) and while the model has its uses, it's also a far cry from SOTA.

So I guess it depends - compared to an Anthropic subscription it kinda sucks, but then again if you have to pay for Anthropic's tokens those are robbery and then DeepSeek looks like a no-brainer alternative.

For the employer those employees cost between 2945 - 7736 EUR per month based on https://kalkulatori.lv/lv/algas-kalkulators (income and social taxes).

So on the lower end that's (1500 USD ~ 1300 EUR) close to half the total expenses of such a developer, on the high end here around 15-20%. That's quite significant, depends on whether their productivity also improves (if that's what the orgs care about).

And we’re not even the country with the worst pay out there, but pay the same for tokens, cause regional pricing isn’t a thing!

Just looked into it, seems like at most they have just 3.2, not 4: https://aws.amazon.com/bedrock/pricing/

Looking around their catalogue more, most of their models seem quite outdated, aside from the OpenAI and Anthropic ones (but those get more expensive). I wouldn't willingly pick Bedrock and would instead throw money at OpenRouter, that has both a bunch of providers, as well as almost any model for you to try.

I bought it for 82 EUR, before the whole ongoing situation.

Now the same spec costs upwards of 290 EUR, about 3.5x the original price and even on Amazon the best prices I can find are upwards of like 210 EUR (2.5x).