redrove

No, it doesn’t seem like it.

I’m sorry but I can’t take a European Commission link seriously about training SOTA LLMs.

Do you have any sources for this? Genuinely interested in reading more.

No it’s not.

You’re generalizing, DACH != the entire EU.

My dream would be a local model that can do, say, 80% of the day to day tasks I need; "how does X Handler connect to Y storage?", "commit that feature, but leave out the bits that relate to billing" etc.

Qwen 3.6 27B can do that today, but setup properly and in a good quant, I run an autoround [0] with weights in int8 and attention heads in f16 on a single RTX 6000 Pro Blackwell Max-Q via vllm with mtp=2 and full context, --max-num-seqs 3, KV in f16, mamba f32.

It would have 99% reliable tool calling

I managed to score 93/100 in tool-eval-bench [1]. For me this is very good already, at least in the pi coding harness I've never had an issue that wasn't auto-fixed in the next turn(s).

the ability to go "this task is beyond my skills" and refer to a Big Boy Online Model in a gigantic datacenter somewhere

This is heavy on the harness engineering side I think, but also quite contrary to the nature of LLMs today. If you figure this out I'd love to know.

[0] https://huggingface.co/Minachist/Qwen3.6-27B-INT8-AutoRound/...

[1] https://github.com/SeraphimSerapis/tool-eval-bench

They’re still debating that, they’ll get back to us soon I’m sure.

I read the entire thing and it seems to me like they started with the conclusion and tried to find proof, like a lot of psychology papers.

Thoroughly unreliable.

It’s about bang for buck. That high a score for 5B params is pretty good, nigh unbelievable a short while ago.

It is my belief that smaller models will get better and better, and even cloud SOTA models will shrink.

Yet another reason the current buildout will feel like the railroads.

Wow, if you you could've just done docker compose up. I guess we'll never know.

So the theory goes. Can you point out one place where this is working out?

Don’t vibe code your auth path folks.

I’m genuinely interested in what you find to be the problem with that?

It’s a landing page.

curl | bash and pentesting go well together

I’m immediately suspicious of cheap chinese crap like this.

$35 for a projector should cause you to raise at least one eyebrow.

Also, as always with “IoT” type devices, they’re best kept in an isolated VLAN with no internet access.

What coding plans are you using that let you use those models specifically and not pay per token? Which provider?

So this is basically vCluster[0] but Rancher branded?

[0] https://github.com/loft-sh/vcluster

cool!

Nothing, but he got to plug his vibe coded startup that he advertises in the about section.

It’s 128b dense model. Good luck getting more than 3t/s out of a mac. It doesn’t matter if it fits or not.

Yes I'm on ChatGPT Pro (OAuth) and I'm trying to use gpt-5.5-xhigh.

That was the issue, 5.4 works just fine.

Support for service: priority (GPT /fast mode) would also be cool!

I keep trying to use dirac-cli with codex and it won't work: Error: Codex API error: Codex API request failed: 400.

Any ideas?

So an Obsidian plugin? Got it.

This is very well put. Thanks for the comment.

It seems to be 160GB at mixed FP4+FP8 precision, FYI. Full FP8 is 250GB+. (B)F16 at around double I would assume.

Gift link: https://economist.com/finance-and-economics/2026/04/21/globa...

I underscored using xhigh + fast mode when saying it’s so much better.

Now with Opus 4.7 of course the “burden” of adjusting reasoning effort has been taken away from you even at the API level.

In my experience people don’t change the thinking level at all.

I’ve been using Codex Pro since they lobotomized Opus 4.6. Codex is so much better, GPT 5.4 xhigh fast is definitely the smartest and fastest model available.

For a while there I had both Opus 4.6 and Codex access and I frequently pitted them against each other, I never once saw Opus come out ahead. Opus was good as a reviewer though, but as an implementer it just felt lazy compared to 5.4 xhigh.

One feature that I haven’t seen discussed that much is how codex has auto-review on tool runs. No longer are you a slave to all or nothing confirmations or endless bugging, it’s such a bad pattern.

Even in a week of heavy duty work and personal use I still haven’t been able to exhaust the usage on the $200 plan.

I’ll probably change my mind when (not IF) OpenAI rug pull, but for spring ‘26, codex is definitely the better deal.

Yes. You could also look at the README.md.

https://omlx.ai/

There is virtually no reason to use Ollama over LM Studio or the myriad of other alternatives.

Ollama is slower and they started out as a shameless llama.cpp ripoff without giving credit and now they “ported” it to Go which means they’re just vibe code translating llama.cpp, bugs included.