Like nobody gets credit for avoiding problems or unnecessary things/complexity altogether. In fact the opposite may happen.

Isn’t this expected? Once it becomes smarter it should event make it unobvious.

its unobvious on purpose, they dont want people to a/b test their jailbreaks

There's virtually an infinite number ways to assess something like this, and a single figure out of context is meaningless.

What's the deprecation schedule? Which financial "context" is it calculated within? A household may benefit from governmental support and profitable, while the aggregate financial situation may or may not be so. What timeline is it calculated on? A 5-10 year window may be unprofitable, while a larger one may be. An even longer one may change numbers completely...

Where do you run DeepSeek?

Discounted pricing is available only at https://platform.deepseek.com. All of OpenRouter providers do not match their pricing at the moment.

I'll also note that the DeepSeek API seems to be really good at caching and their cached input price is more heavily discounted than most providers at $0.003625 (vs. $0.435 for input cache misses). So, it's hard to spend a lot of money fast with DeepSeek.

I was concerned I would need to do something specific in my dumb agent harness to make caching effective, since I'd read Anthropic's reason for forcing people to use Claude Code in order to use the rolling token usage limits on a subscription was because they could control cache behavior more effectively, but DeepSeek seems to be able to handle caching very effectively for raw API calls.

It's not discounted pricing anymore, it's the regular pricing.

I used the native DeepSeek API at deepseek.com. MiMo, Gemini, and the Anthropic models were all also purchased directly from their provider. The other models in the bench were either on OpenRouter or self-hosted.

That’s interesting! As I’m researching this (for my family) I have questions :) - have you tried entomol? How does it compare? - is it advisable with for young kids, if it’s irritant? - how long does it stay on? - what do you do when you get so many ticks on yourself? Watch if you get spots after 30 days?

Entomol is hydrogenated catmint oil

Never tried Entomol. But I've found any of the essential oils simply don't work more than 10-20mins after application. I'd be very skeptical, but I'll try it if I can find an EPA/FDA approved spray.

Picaridin is safe for everyone, but follow all label instructions. Also use the 20% concentration. I find that 10%-15% is about as effective as DEET, 20% is about twice as effective (in my completely uncontrolled, unscientific, testing).

> is it advisable with for young kids

I can't comment on the safety factor, but to me it smells like cotton candy, so your kids will probably be more willing to put it on than DEET. (I don't know if this is how the chemical inherently smells or just the brand that I have though)

> if it’s irritant?

You definitely don't want to get it in your eyes or mouth, but I have pretty sensitive skin and I've never felt anything when putting it on before, so it's not that irritating in my experience.

> how long does it stay on?

I seem to remember that the bottle says 12 hours, which is a bit optimistic, but not too far off. I think that I've only ever needed to reapply it once, whereas I've sometimes had to reapply DEET multiple times in a day.

+1 - MCP is a transient effect of current models’ limitations. But the “when” is relevant

I wonder what’s the motivation though. Subscriptions? At a minimum they could limit based on recency. It’s “news” after all so having a number of days of delay for archival would solve most issues, I assume.

I think it’s a cost/opportunity tradeoff at best with any agreement, regardless. The rest of the contract may make it difficult to impossible to do anything about it, starting with basic arbitration clauses and ending in a ton of other provisions that can make any legal action futile. I doubt there’s much room to negotiate too.

Given that all labs need to diversify to become profitable, they’ll end up competing with their customers and theres nothing that exposes a business more than having AI offload every job function for every account, every mail etc.

Assuming this won’t be an issue is naive at best.

This is more like the popular lay take that are more or less confused about the meaning and implications of the theorems. The fact is that the implications are real yet more nuanced than this - something the article touched on.

Hilbert’s program was to reduce math to a finite set of axioms and was indeed derailed by incompleteness theorems(Gödel) and undecidability (Church, Turing).

Formalizing math with automated theorem provers has little to do with the goal of Hilbert program. Also they aren’t related to the foundational research that it entailed.

Also, as the article mentions, the implications as well as Gödel was largely misunderstood.