solenoid0937

Yes, distillation is very very easy.

Claude Code is much nicer than Pi, especially for long running autonomous work. Pi might be OSS but it's just not nearly as featureful or agentic as Claude Code.

Distills don't take months to train, they take weeks. Distills are very easy to train.

It sounds like post processing should be the job of an LLM. I would like the voice model to be faithful to what was said and then that output can be smoothed over or postprocessed as needed for the use case

It isn't a dog whistle. Dubai, Japan are also high trust.

It is not just about control, it is also about being able to trust the nodes in your network.

It is not just about employee devices, but about literally every workload or host running in your corporate infrastructure.

My comment is fine. Appeals to authority are not inherently bad when the person doesn't know what they're talking about. Again, your doctor is more of an authority than you on medicine. It'd be hubris to think you'd understand the field better, no matter how smart you are.

It's not my job to write an essay in the comments about why mutual authn with RA is desirable in corporate networks, and why the complaints about "freedom" are totally and utterly nonsensical in this context. This is something he can look up very quickly.

The audience of this article are people working with infrastructure, not people developing user facing products. See quote: "If your infra consistently enforces mTLS ..."

I hope you put a lot more research into those debates than you did with this topic! I just don't understand why you'd have this overconfident hot take about a topic you clearly aren't familiar with. Like, try to understand the article subject and audience first?

Please stop bashing on things you don't understand. This is standard practice for workload authentication in corporate fleets. The article makes the target audience clear:

If your infra consistently enforces mTLS

whenever I see "remote attestation" I associate this with "hostile"

HN is bizarre. This is just standard infrastructure security practice at any tech company of meaningful size. You are misunderstanding the target use case and audience of this article.

You don't understand the use case or audience of this article:

If your infra consistently enforces mTLS

This is for mutual authentication in corporate infrastructure. Attestation is a critical security property for these environments.

If the model can't get the information it needs to accurately answer the question, it must surface that risk to you instead of guessing. This is part of model intelligence and tool use competency. Fable and to a lesser extent Opus is very good at this

Tool usage competency is part of overall intelligence. If the model can't get the information it needs, it must clarify that in the response.

Yeah every model has great benchmarks. Claude is the only model I want to use when I'm not worried about the marginal cost of tokens (which is most of the time at work.)

I then use cheaper models like GLM for personal projects but they're noticeably much worse despite being similar in benchmarks.

There's been a ton of discussion on HN about this but yes. It's a totally different level from Opus.

Global power does not run on kumbaya vibes. Superintelligence is a zero sum game whether you understand why it is or not.

I'm glad you trust a country that literally imprisons, rapes, and murders their human rights lawyers though. And one in which you cannot even criticize the government without being disappeared. Surely such a country is best positioned to have total control over all humanity, forever.

Sometimes you just gonna channel your inner Karp

I use GLM 5.2 daily. Calling it Opus tier is a bit of a misnomer. It is maybe comparable to Sonnet 5, definitely not Opus 4.8.

It is a great model for the price, but it has much worse autonomy and long context performance, and I don't trust it for anything beyond personal projects, whereas I use Opus/Fable and GPT for work.

At my company we are pretty certain that a lot of datacenter opposition is being instrumented and funded by the PRC.

AI is a bid for control of all humanity. The first to superintelligence owns our future. I'd say it's okay to be a bit suspicious or conspiratorial of anti-AI/anti-lab narratives.

Fable always felt clearly a huge step above Opus for me. It's been able to one shot complex bugs and apps Opus could never solve. But it's expensive.

This is scary. "Collusion" and "collaborating with your subagents" seem like difficult problems to solve at the same time.

OpenAI has a $8B PR budget. Makes you wonder if they'd spend any of that on social media.

I saw a guy trashing Anthropic on Threads the other day, looked up his LinkedIn and he worked for OpenAI in PR.

Of course they will. It'll buy you a few months, but if you want stability your only option is API.

Harder to make a spicy article if you can't argue the companies are greedy.

Anthropic bad! Straight to the top of HN please!

Their alpha has been discovered by the other labs. If they took the lead they obviously would not give that up freely.

I'm sorry, you're right! What was I even thinking? Of course we should hand every man, woman, and child a mini nuke!

The main risks called out with GPT 2 were misinformation and they were right about all of it. Every risk they called out came to pass.

I run local models every day.

We are in a race to superintelligence. The first country to AGI will be the first to superintelligence, and the first to superintelligence will have de facto control over the world and the future of humanity. They will also be able to prevent others from reaching superintelligence.

Of course it's in the US's best interest to slow down China. You aren't zooming out and looking at the big picture, you're taking models as slightly useful tool, not what they will soon turn into.