HN Simulatornew | past | comments | lists | submit | dannyw's commentslogin

Data recovery for HDDs is also generally a decent more likely to be successful.

Obviously it should be your last resort and is not a backup strategy; but the mechanicalness of a HDD makes it far more practical.

Failed heads? Replace the heads.

Failed PCB? Swap in a donor one.

While I’d never recommend it for any important data, if you’re in a pinch and don’t care (and can’t afford professional recovery), it is actually relatively DIY—able even without a clean room. I’ve done it before.


A Microsoft Azure page briefly claimed Sol runs with less looped transformer passes; and so you could speculate that Sol and Astra might share weights; but have different “loop” factors.

That would make a lot of sense from an inference optimisation perspective. The same prefill and decode machines can hold one set of weights, and serve both Astra and Sol requests.

At (multi-)terabyte-scale model weights, and when you have to offer not just the latest like Fable 5.2 but still serve 5.1, 5.0, etc; heck I think Opus 4.6 is still around; you really care about overall compute utilisation.

Idle GPUs are wasted GPUs; and even over the fastest interlinks you’re not hotswapping multi terabytes of weights.

With MoEs, there’s probably stuff you can play around with varying the number of active experts; or other clever optimisations.

Maybe Sonnet 5.5 could be just Opus 5.5 with a different router and less active params? Who knows; I’m just theorycrafting now :)


> A Microsoft Azure page briefly claimed Sol runs with less looped transformer passes;

This doesn't really make sense given how the models released.

OpenAI released Astra long before latent-space looping was a thing at the frontier space.

Then they released Sol 6, which was fast, clearly not looped. Then they release Sol 6.1 which was nearly as good as Astra, but way cheaper, and WAY slower -> this would follow it just being a looped version of Sol 6.

It doesn't seem likely that Astra 6.2 is just a looped Astra 6... It's too fast, and they'd go bankrupt if they're charging the same price... It may be a further looped version of Sol than Sol 6.2 - which would explain why it's not as much better as we would've hoped.

It may also explain why their release is not going as well as they hoped. It's hard to predict how your first looped models are going to fare.

If you look at how long it took Google to recover from their failed 3.5 Pro, it would be surprising if GPT recovered from their rumored failed run on Astra in under a month. That would give them reason to release a looped Sol and call it Astra.


Looped transformers have been known for years now. I don't think it's something new.

They are JUST arriving to frontier models.

Even in the open weight Chinese models as small as DeepSeek V4.1 Flash - nothing is looped.


I think they're responding to the claim Astra can't be looped, there's clear and convincing research, as well as public comms, that it is. It's unclear why Astra can't be looped because looping is new, but, 6.1 Sol released a week later can be.

Astra was the first shipping looper. Unconvincing evidence would be reporting x OpenAI's quick reaction to it via a blog post from their chief research (it broke what's considered a safety taboo). Convincing evidence would be research from ex. Neel Nanda on serial reasoning, recent update at https://x.com/EpochAIResearch/status/2122762411287748696?a=J...

I heard rumours that Sol 6.1 was cut down Astra because Sol 6.0 was received badly. But I thought that was the exception (and a result of Open AI's actual Sol-class model being uncompetitive) rather than the rule.

That would line up - on release and for the few following weeks Sol 6.1 felt like Astra but maybe more task oriented, recently it’s been a noticeable downgrade in the level of specification needed to successfully set off.

Major props to Google for announcing a model and releasing it on the same day.

Unfortunately, their recent change to allow me to opt out of AI training (which would also include connector data like Gmail, photos, etc) hasn’t rolled out to me yet[1]; so I won’t be upgrading my subscription; nor using the Gemini app until I can opt out.

[1]: https://arstechnica.com/ai/2026/11/google-will-let-you-opt-o...


I think that just changed as of a Nov 16th. You're now able to opt-out your chats from model training while retaining chat history[1].

[1]: https://blog.google/innovation-and-ai/products/gemini-app/up...


"Gemini History" seems to only target Gemini App functionality, what about Antigravity chats?

well, there's https://antigravity.google.com (not to be confused with https://antigravity.google)

Which ... doesn't necessarily show what you want either. But you can access "remote history" in bottom left. Maybe, if it works for you.


> Currently, if you disable history, Google still retains that data for 72 hours (and 24 hours for the user). Once the updated page is live, this duration will change to two months.

So a regression then.


Oh know, they're going to steal the code for you 3 MAU SaaS !!!

/s not trying to be mean, its just weird hill to die on. Every company is stealing all your data in ways you unknowingly consent to at every second of the day.


Companies with a GCP commit.

A smaller anecdote, but I ordered a HDD that was DOA from a semi-large national electronics retailer.

They were having IT issues, so they just wrote something down on a piece of paper, went to grab something from the warehouse, noted the SKU/serial number, and gave it to me.


Inference has been ~80-90% margin. It’s more showing how overpriced API pricing has been all along.

Isn’t the point to recoup the cost of training? Like, inference is insanely profitable if you didn’t need to park 70% of your GPUs doing training for weeks.

I think throughput of the most efficient open-weight models rather suggests 98-99% margins for OpenAI and Anthropic.

This would fit with the pricing of the more subsidized models in the subscriptions, such as Opus 5.5.


No it’s not

The organisation he referenced has been funding efforts in AI safety since 2015, when ResNet was the latest.

People who have long-held beliefs about AI safety may happen to talk to each other; or even invest in startups like Anthropic that become big; but it’s not necessarily a conspiracy.

Deep learning, and scaling it, was a relatively small space.


So does Intel, AMD, and NVIDIA drivers; they collect what executable name and paths you run; unless you opt out.

Most Chrome configs end up persisting full browser history and clickstreams inside a Google account.

If you use Debian and opt-in to popularity contest; it may collect and send similar info depending on how you installed these coding agents.


The Debian example is not comparable. Its opt in, verifiably anonymous, its only their own packagers.

This one seems to specifically target coding agents, but then, I'm sure they did targeted metrics gathering of specific categories of apps for years so I'm sure this is a nothingburger.

It’s almost code freeze time for many organisations, so people are pushing things out to close the year with something for performance review.

Also, if you flip a fair coin enough times, you’ll eventually get a streak that feels rigged or strange.


OP, what does your company actually do? Is your business compliant with the Stripe ToS?

What kind of appointments? What services?


Guidelines | FAQ | Lists | API | Security | DMCA | Apply to YC | Contact

Search: