rw2

Sure, China will build a lead in robotics, drones and weapon tech.

They will invade Taiwan, US defends, WW3.

If they don't have a military/tech lead, the status quo remains.

Totally agree with all of this. Most of the problems we solve with AI are not one shot, all the bench marks also miss the human components when testing. Thinking alongside with a human and coming to a solution is what Fable does better than most other models.

The amount of scientific talent in China is astounding, also it's easier to be a fast follower than a innovator.

Instead of limiting models and debating ethics like Anthropic, the edge lab should focus on lengthening the lead on China.

The 2027 Chinese model could be one that beats the US.

There are now 3 tiers of competition:

-Fable + Gpt 5.6 Sol

-Opus + Gpt 5.6 Terra + Grok 4.5 + Muse Spark 1.1

-Open Chinese models: GLM + et family

The economics is on the Fable tier people are willing to spend a lot on it and on the Open tier you have to give it away to drive usage. The bottom tiers are also getting more and more competitive.

Engineers to crawl that content cost money too

The papers can make it free, they can just build a convenient search/retrieval API layer so Open AI and Anthropic don't pay 1-2 engineers a year at 1-2m to crawl and index that info for training.

The AI services have an option then to pay for this service, support a open service, or write their own crawler. I think if every open AI request didn't just do a web search but a more targeted arXiv search the results would be better.

They are so consistently wrong and out of touch with reality they are the Cramer of publications.

Should charge AI for training on top of it or get them to donate. A small amount can fund them easily.

The use of the "cheaper models" in big AI companies are next to useless as they don't even score as well as the open/super cheap Chinese models. Only the frontier big models like Fable and Opus have value.

This is making the case for Anthropic KYC for US citizens. No one would allow their accounts to do this if they were on the hook for it from the US government.

This is a good thing:

A. This brings responsibilities on using these super powerful models like fable and mythos.

B. Distillation attacks can be avoided as someone who does it can be perma banned

C. Brings value to having an American passport, AI benefits all the world equally but it's built on American talent and capital. There should be some value to people who are not shareholders.

Good guide but I think the product market fit portion of a startup is so key that you need no other skills except that to make a good startup. AI won't help you with that portion, only in depth knowledge of a industry or natural product intuition will.

Who knows, maybe an AI ideated and AI created product will be the best app of 2026.

This makes much more sense. Whatever criticism for trump he's realist who does want the US to win the AI wars.

Anthropic's own marketing and urban legends spawned by them is to blame too. They built up too much BS around mythos and project glasswing

Not really, they are not even as good as opus 4.7

Claude Fable is a insane improvement that is not reflected in any benchmarks that are currently out because the improvement are on the hardest problems.

They didn't put Opus 4.5 on the model card to compare

Having used it compared to Nano Banana:

-The latency is still too high, lower than 10 seconds for nano banana and around 25 seconds for GPT image 1.5

-The quality is higher but not a jump like previous google models to Nano Banana Pro. Nano banana pro is still at least equivalently good or better in my opinion.

Gemini 3 in antigravity is significantly better than Claude code with either Opus or Sonnet that I struggle to see how they can compete. And I'm someone with the 100 dollar/month plan.

I can't even use Opus for a day before it runs out before. This will make it better but Antigravity has way better UI and also bug solving.

Gemini 3 in antigravity is amazing

I believe that the fact that Chinese models are beating the crap of of Llama means it's a huge no.

The way I would do it is I release a open source code without any spyware then prebuilt dmgs that contained it. Thus the lazy people would all get it if they dont self compile. As compiling a new thing takes another 1/2 hour or more for most people, they would be too lazy to do it.

I love the product concept but the fact this person has an almost empty github and suddenly launches an app that can easily be spyware concerns me a lot :). A lot of security concerns with password etc.

I think the future with AI has no room for Palantir.

Basically, their entire premise is go into some place, collect all the data and build models that are useful on top that data.

Those models are now pretty much useless in the age of LLMs as the LLMs are so powerful you dont need custom models to predict behavior anymore. The new meta in this space is probably someone taking all the data into some db and using LLM on it either through training or interpretation.

Disagree, almost of all the accomplishments in humanity are driven by people like Steve Jobs not Woz. Elon Musk could be said to be a second iteration, a technical person extremely good at sales that can pursue and sell a vision.

There are a lot of people who want to be happy. Let them be happy, but it's the relentless builders/dreamers who pushes through the entire journey of getting a product out there to the people.

Claude code is a poorer version of aider or cline in VScode. I have better results using them than using claude code alone.

non AI IDEs are going to have a hard time in the future.

I tried this vs Cline/Aider/Lovable.

For full stack, I would say Lovable is the best, for complexity, I think Cline is the best.

Claude Code with the several request I gave it just produced code that didn't run. I think it has a long way to go.

Is not having to build a dockers image worth $100 a month? I do find server setup to be a pain but I think if I will use a model for a year, I can take the time(3-4 hours) to set it up. Only with constant switching of models would I use a service like this.

I never set up bigger models like LLAMA on servers. Other hacker news people can chime in.

Great product, first of all. I can really see a use for it. Are you afraid that this is too easy to clone?

Someone with speechify: https://speechify.com/

And who wants to write a spotify API write code can do this.