wasabi991011

How would that help with the raised issue?

That's my biggest issue with it personally, as an academic. Way too much unnecessary indirection that makes it harder to read and understand the assumptions and parameters going into the code, and unnecessary defensive programming for code that is just going to be run a few different ways with strictly controlled inputs.

The documentation on swarm has a short comparison with the other options: https://antigravity.google/docs/swarm

When's the last time you checked what is offered by the plus subscription? It's been a while now that it's the equivalent to ChatGPT Go (on price and offer), giving only increased limits but no difference in model access compared to free.

From what I've read, it seems like the whole point of the RTR is to not be like that.

Can you explain to me what The Clearing House is as an organization, and how many participate in their RTP? I have tried to look this up but can only find vague promotional material by banks.

Thanks for sharing that link btw, it was great! You should post it on HN for more people to see.

That is not true. Especially since TFA clearly dates the invention of parchment paper to 1847, nearly a century before the synthesis of PFAS.

a quick google search says yes

I had a research question in my area of expertise. Tried it with ChatGPT, Claude, and Gemini. I think all three could have solved the question, but ChatGPT was an unreadable mess with too many sections and headings, Claude got stuck thinking and never produced an output. Gemini gave succinct explanation that I think was similar to ChatGPT conceptually but that I was able to follow and verify. The answer was actually fairly trivial to understand; I just hadn't gone at the problem from that perspective.

I know it is common for people to try AI in domains they aren't experts in, but it is also common for people to use AI in domains they are experts in.

I agree, I don't think Plus is a good value proposition

I didn't ignore anything. I chastised for the missing information and gave a potential reason (which could totally be wrong for who I was replying to, which I acknowledge by saying if).

Thanks for sharing your experience though, it's a more substantive piece of information than who I was replying to.

Still not saying which paid plan. I don't really understand why you think that doesn't matter. Google is not the only company to have different model availability for different plan tiers.

Argon has also been publicly available for the same amount of time for Pro users.

I don't understand why HN is so surprised to have an announcement for model existence and a follow-up announcement for model availability. It's really not that unusual, as you say Anthropic has done the same, probably others too (though I can't remember)

What does /boost do?

And do you have advice on how best to use Gemini models? It seems like my usual strategy of iterating on plan md file -> implement based on plan doesn't seem to work, 3.8 tends to rewrite the plan too often removing crucial parts when iterating.

Disingenuous to not mention what plan you are on.

If you are talking about free versions, OpenAI and Anthropic also have a difference between free models and models available to 25$-tier plans.

Note this is also true of OpenAI and Anthropic too (except Anthropic doesn't have a tier between free and all-models-available)

Same as the ChatGPT Go: higher usage limits, more image and pdf uploads, more tools

Not really[1]. I had access to Argon within an hour of the release announcement.

The Google AI Plus tier does not give better model access than free, it is strictly for usage limits. Same thing as ChatGPT Go (which currently only has access to GPT-5.6 Luna, same as the free tier too.)

[1] For consumers at least. IIRC google business/entreprise accounts have worse model access even in the business/entreprise version of Google AI Pro.

Finished reading this book not too long ago, one of my top three textbooks. An absolute joy to read full on great insight.

Not sure how it feels to a beginner, but to someone with a good deal of familiarity in the field it put everything in such a nice tight conceptual package.

Great stuff. Two questions:

- Are there any issues with vocabulary drift? I know the LLM field in 2017 is not super different, but is there no atoms from 2017 that ended up being irrelevant to 2026 and/or which are just not possible to combine with more recent atoms?

- What does the cost of this project look like? (Ignoring the research iterations, just the costs relevant to the final product).

Can you give some sort of breakdown of the costs involved to make this project? Once the methodology had stabilized.

No, the medRxiv preprint is linked at the bottom of the page. It is freely accessible.

Edit: I guess it depends on interpretation, the press release does start with "The study, available to clinically eligible Biograph Black tier members," which makes no sense to me. Maybe they mean enrolling as a patient in the study was only available to paying members? This would fit with the study population description in the preprint.

Thanks for the responses, I'll check it out!

The starting states being at different points in the task and having success-guided starting state sampling seem like really neat ideas. I haven't kept up with RL in a while, how novel is this?

Ah I see now, thank you.

I'm not sure what you mean, Graham's number is trivially computable since it is a natural number (finite).

Graham's number is not a sequence, so it doesn't matter that the BB sequence grows faster than any computable sequence.

Why does the directionality matter? The initial sentence was untrustworthy, so a resentencing is necessary. Discussing the directionality of bias would just introduce another point of failure and more inefficiency in the system: maybe you believe the direction of bias is clear here, but I'm in many cases it might be less clear and so you would necessarily have to get lawyers involved to make arguments on each side.

It's not a misrepresentation: 13 out of their 21 posts are about Israel.

I don't know why you brought comments into it, the most common reading of "posts" is "submissions", not "submissions and comments".

Sure, your students may very well spend the little free time they have learning maths.

But are any of them going to be able to do any significant amount of studying maths without being homeless?