Hacker News: pants2

New comment by pants2 in "Claude Opus 4.8"

pants2 — Thu, 28 May 2026 17:55:53 +0000

Yes, because of the limits on tok/s, and you have to compare apples to apples, not Gemma 27B to Opus 4.7.

New comment by pants2 in "Claude Opus 4.8"

pants2 — Thu, 28 May 2026 17:41:55 +0000

Running them locally is cool and has privacy/autonomy benefits, but you can't really make a value case for it. Guaranteed if you run the math you will never run enough inference to pay off your hardware vs buying tokens. Last time I ran the math on my MBP I'd have to run inference 24 hours a day for 5+ years to pay off the cost of my MBP, not accounting for electricity costs.

New comment by pants2 in "Claude Opus 4.8"

pants2 — Thu, 28 May 2026 17:36:38 +0000

Yes, I think this has become their competitive edge to stay relevant and retain customers. If a lab falls behind the frontier for too long, they will lose customers to other models. Google, DeepSeek, and XAI have all released frontier models in the past, but they fall behind and people lose interest.

New comment by pants2 in "Claude Opus 4.8"

pants2 — Thu, 28 May 2026 17:31:17 +0000

Polymarket says not likely until the end of June. Maybe some money to be made?

https://polymarket.com/event/gpt-5pt6-released-by

New comment by pants2 in "Claude Opus 4.8"

pants2 — Thu, 28 May 2026 17:23:09 +0000

The Chinese models are only cheap on subsidized Chinese hosting. I have yet to find a USA-hosted Chinese model with a very clear value advantage over US models.

New comment by pants2 in "What Apple and Google are doing to push notifications"

pants2 — Wed, 27 May 2026 22:00:36 +0000

I would love if Apple enforced that rule, but they certainly don't

New comment by pants2 in "What Apple and Google are doing to push notifications"

pants2 — Wed, 27 May 2026 21:59:34 +0000

I definitely run all my emails through an LLM filter and wish I could do the same for push notifications!

New comment by pants2 in "What Apple and Google are doing to push notifications"

pants2 — Wed, 27 May 2026 20:31:00 +0000

The biggest problem are apps that do both. For example, I want Uber to notify me when my driver has arrived, but I don't want it to notify me when they have a special 10% discount on my next 5 rides. It's not straightforward to block one but not the other.

New comment by pants2 in "Search engines alternatives now that Google isn't Google anymore"

pants2 — Mon, 25 May 2026 14:35:29 +0000

I have been a Kagi subscriber for years but I do increasingly find myself using Google. It's not good at local searches or news searches. It's also not good at showing quick context like Google's knowledge graph, especially with images. Finally it's considerably slower.

Kagi is better for research and knowledge work, Google is still better for quick lookups.

New comment by pants2 in "Waymo pauses Atlanta service as its robotaxis keep driving into floods"

pants2 — Mon, 25 May 2026 04:39:47 +0000

haha you got me there, can hardly even consider those as roads!

New comment by pants2 in "Waymo pauses Atlanta service as its robotaxis keep driving into floods"

pants2 — Fri, 22 May 2026 14:54:39 +0000

They have done a lot of testing in Pittsburgh which has some of the craziest roads and intersections anywhere, so I'd assume yes

New comment by pants2 in "Starship's Twelfth Flight Test"

pants2 — Wed, 20 May 2026 23:27:01 +0000

That would take ages!

New comment by pants2 in "Meta blocks human rights accounts from reaching audiences in Saudi Arabia, UAE"

pants2 — Wed, 20 May 2026 15:31:06 +0000

In a just world all companies would be taxed on their overall impact and not just revenue. Coca Cola would be taxed for their contribution to obesity and plastic waste. Exxon would be taxed for their emissions. Meta would be taxed for its harmful impacts on society and childhood development.

New comment by pants2 in "Google changes its search box"

pants2 — Wed, 20 May 2026 02:01:23 +0000

Yeah the Google AI results are more dangerous than ChatGPT, not only because it uses a smaller model but because Google's knowledge graph used to deliver very accurate and authoritative information but now that's been replaced by a stochastic system in the same place, so people are used to trusting it.

New comment by pants2 in "AI license plate cameras tore this town apart and led to a state of emergency"

pants2 — Sun, 17 May 2026 15:41:41 +0000

> Last month, Mayor Carmella Mantello, flanked by officers in blue, accused the city council of “defunding” the police and declared a state of emergency to keep the cameras running, a designation usually reserved for floods and blizzards. “I will not put our city in jeopardy and take these cameras away,” she said. The lengths that this Mayor goes to to keep the cameras on raises suspicion.

New comment by pants2 in "MacBook Neo Deep Dive: Benchmarks, Wafer Economics, and the 8GB Gamble"

pants2 — Thu, 14 May 2026 15:29:42 +0000

I used an M1 Air for developing iPhone Apps for years and it worked great. Not "wow" fast but I never had a complaint about it.

New comment by pants2 in "ZachXBT traces $19M crypto social engineering scheme by a teenager"

pants2 — Wed, 13 May 2026 21:00:40 +0000

Criminals posting evidence of their crimes online certainly makes the prosecutor's job easy.

I wonder what's going through his head though. Did he just not expect to get charged? Or just wanted to live life to it's fullest knowing he's going to prison anyway?

New comment by pants2 in "Googlebook"

pants2 — Tue, 12 May 2026 19:05:43 +0000

You might be surprised how good cloud gaming has gotten. I play AAA games at max settings on my MacBook Pro through GeForce Now, and with fiber internet it's nearly indistinguishable from native.

New comment by pants2 in "Gemini API File Search is now multimodal"

pants2 — Mon, 11 May 2026 20:55:28 +0000

I think the real crux of the moat is model intelligence. I'd bet that most of the money being spent on inference is on the top few models (today Opus-4.7 and GPT-5.5) from people and companies that benefit from using the best models.

Truly the main moat that OAI/Anthropic have is being 6 months months ahead of the competition in performance, which might be indefinite if the competition is just distilling their models (China) or takes many months between releases (Google).

Once you look passed the frontier of performance, it's just a race to the bottom on inference costs because there's at least 5 companies with equivalent open models at that level.

New comment by pants2 in "Gemini API File Search is now multimodal"

pants2 — Mon, 11 May 2026 02:47:45 +0000

- Right, R1 affected markets because the market originally believed your theory, but it doesn't any more, which is why V4 didn't move markets at all.

- Sure, you can use US infra providers. Together.ai is a good US provider but then it's 15X more expensive than DeepSeek's Chinese-subsidized pricing. It's really not that attractive at that price point. Anthropic and OpenAI are focused on larger models, but Grok 4.3[1] is smarter and significantly faster + cheaper than DS4[2] and by a wide margin.

- DeepSeek has a Claude-compatible messages API, but that's trivial. Anthropic has a massive API platform with things like Sessions, Files, and Agents[3]. None of those are available on DeepSeek.

1. https://artificialanalysis.ai/models/grok-4-3

2. https://artificialanalysis.ai/models/deepseek-v4-pro

3. https://platform.claude.com/docs/en/api/overview