I believe the future lies somewhere in between. I'm working on a hybrid application to reduce our company's token consumption. It runs on our data center's computing infrastructure and on laptops in our community. You might be interested; you can check out the code if you're interested: https://github.com/vfalbor/hibrid
I used to use wikiloc, but most of the things that offer which were the most interesting things were by paying, so I think that it could be some opportunity for using these maps and vibe coding for creating something spectacular!
There is one email-newsletter (Top daily news from HN commented) that works from several months ago for HN that I'm using, you should check it: https://tokenstree.eu
I have tried it and I use it. I think it's going to become the standard way of operating, especially when they start charging us an API fee, which is supposedly the real cost. But of course, with how much they charge for the token and depending on the model, there are so many factors that I think the future is heading towards local models. I believe there are good models out there, and the key is the concept of "pruning," where you select the layers that interest you most and try to reduce the hardware cost of these types of models. The Qwen and Gemma models have been discussed here, but Kimi, which is a fairly powerful model with an efficient pruning system, could be your perfect free co-pilot in terms of coding, and could coexist with the more powerful Opus or Gemini models. The key concept is skills that make this process transparent.
If that were the case, it would be reasonable to expect that companies like OpenAI or Anthropic, which are heavily indebted, would lose part of their business model, not because their models are bad, but because others will be cheaper and not as bad.
This is a very interesting comment. Companies like OpenAI or ChatGPT sell hardware hidden in tokens, and the token is different for each company depending on the tokenizer. The concern is this: when you have an Opus 4.7, Sonnet, or GPT 5X with an Nvidia H100 or H200 GPU, what will happen to this cost when, if not Nvidia, another Chinese company enters the market and starts running these models? The point here is that as long as Nvidia is the provider, and limits access to the machines and the number of data centers is also limited, these companies can be worth whatever they want. But the moment this starts to expand, the value will surely decline, because what you're selling isn't the model itself, which is ultimately just a 1 TB file that you have replicated across machines. What you're selling is access to a software program on a specialized machine. As long as you control the resource, which in this case is that machine, you'll have value. The moment other machine manufacturers enter the market, your value will decrease.
A few weeks ago, I saw a documentary about how inefficient and unstable these types of trucks were. It was necessary to redesign the cab's aerodynamics to achieve substantial fuel savings in these vehicles, which are inherently fuel-intensive.
I think LAN parties made sense in a context where internet speeds weren't what they are today, where pirated software and games were distributed in these environments, and where you could get together in groups to play CS, Starcraft, or Age of Empires. But nowadays, with internet access and resources so widespread, and peer-to-peer networks offering countless more than eMule, they've lost their original purpose. Perhaps a pivotal shift for these events is needed, focusing more on building social networks than on the simple idea of eating pizza and sleeping on inflatable mattresses on the floor.
Yep, actually, is a mixture that works. I actually run for my day to day, and I can save tokens, maybe not that I will expected, but it works, you can try if you wish https://translation.tokenstree.com.
Yes, it is. In fact, I made a small application to reduce the token consumption for translating from one language to another, and I even invented a language called Tokinensis, which is a mix of different languages, and I ran my own tests with savings of 30%. Chinese is amazing because they encapsulate a ton of information in a single symbol, so you can save a ton of tokens.
This is perfectly legitimate. It's something I've been denouncing day after day. Company X charges you 10dolar per token, while company Y charges you 7dolar, yet company X is cheaper because of the tokenizer they use. The token consumption depends on the tokenizer, and companies create tokenizers using standard algorithms like BPE. But they're charging for hardware access, and the system can be biased to the point that if you speak in English, you consume 17% less than if your prompt is written in Spanish, or even if you write with Chinese characters, you'll significantly reduce your token consumption compared to English speakers. I've written about this several times on HN, but for whatever reason, every time I mention it, they flag my post.
PhD in Distributed Computing from the University of Santiago de Compostela (USC)
Author of https://github.com/vfalbor and tokenstree.com.