dcchambers

I would really love to be able to use these Gemini models in Opencode or Pi with my existing Google AI Pro subscription.

Given the choice between a walled garden like Apple's app store or this, give me something open and more risky any day of the week.

It's easy enough for Omarchy folks to add "verified" or "trusted" developers down the line.

"If you own someone $340, that's your problem. If you owe someone $340,000,000, that's there problem."

Might as well be a trillion dollars as it's never getting paid.

Testing capacity?

Technically speaking everything costs money to compile in the form of electricity. This is just...inefficient compilation.

Tokens are just a type of currency. Stripe is a middleman for currency.

Using something like OpenRouter (or any of the AI Gateways) is better than tying yourself to one LLM provider that can rug pull on pricing or change models in a way you don't like.

The value is in the network effect I think. OpenRouter is popular and has a good head start over anything Stripe could build internally.

This latest DeepSeek is almost at the "too cheap to meter" level. That's going to be a larger unlock than models like Fable/Mythos that are way too expensive to justify, IMO.

What secret sauce do they have?

> Why is every company seemingly trying to re-implement agents themselves rather than making their product work with the existing ones?

Why does any company do anything? Money, of course.

It's 100% okay for people to write in their own books, and it's okay for them to donate or sell them after the fact as long as they don't hide the fact they are written in.

No one is forcing you to buy used books with writing in them.

It's incredible that each model has it's own style that remains relatively consistent throughout all of the different generated examples.

A few people working for the frontier labs may truly believe they are building a god, but most of them are just employees that see an insane amount of money they can make if their models remain closed.

Most of them aren't worried about AI safety, politics, religion, etc. It's really not that deep. They just want to get rich.

There's nothing wrong with that, but let's call a spade a spade.

I think hardware like this is the future for LLM-providers once we reach a point where the models aren't advancing much any more. You could argue we're close now.

The hyperscalers like AWS will made great use of these to serve up models that will be relevant for several years. But right now, we're still seeing significant bumps in model quality every couple of months - especially with open-weight models like Deepseek/Kimi/GLM.

Until that point, though, I don't see how this is ever going to be cost effective vs general purpose hardware.

I also think we'll see miniature versions of this baked into mobile hardware for super fast and efficient on-device LLMs.

Most people IN TECH still don't realize how much hardware prices are going to continue to increase, much less the 95% of the population that's completely unplugged from what's going on with AI-driven hardware demand.

All of these price increases are going to get passed down to consumers eventually via increased prices.

Being unable to use this with zero data retention makes this feel like a non-starter for most enterprise customers.

IME Opus 4.8 (and 4.7) is often a downgrade from 4.6. I find that it tends to overthink and overcomplicate things.

Hobbyist computing is dead for the foreseeable future. These prices are untenable.

I think two things happened:

1. The sheer number of tokens that a coding agent can use flipped the math upside down on this equation. If you use the most expensive model for everything those costs quickly become untenable, even for software companies.

2. We realized many of the coding problems we're solving aren't incredibly difficult.

Probably some combination of: Anthropic is heavily invested in the Rust ecosystem and they want their core tools to be built on Rust. More Rust developers. More Rust training data so LLMs write better Rust code than Zig code. Advertisement for Claude Code doing major work on a high profile open source project.

In 2026 the answer is "job security"

Their pricing has always been a challenge, but I think the biggest issue is that they went way too wide early on in an effort to capture the market. Their product pool was a mile wide but only an inch deep. People quickly realized many of the features they were promised were very rough around the edges or a limited POC. Their UX is pretty unintuitive as well, IMO.

The company is fascinating. The "everything is open" nature of how they build the business is an incredible value for people that want to see how the sausage is made...but it hasn't ultimately lead to a better product.

Incredible. Mutually assured destruction.

The next five years are going to be truly WILD in the software world.

Air-gapped systems are gonna be huge.

Obsidian.

Checks all of your boxes:

   - Backed up to cloud 
   - Has a mobile app that is easy to make small changes 
   - Simple file format like markdown 
   - Option to export data out as a backup  (it's just markdown files, you don't even need to export, just copy them. Or export to PDF)
   - Allow uploading arbitrary files like PDFs, images, receipts etc...  (https://obsidian.md/help/file-formats)
   - Nice to have: support for inline tables of data with simple calculations/sorting

Google Docs doesn't do classic "wiki" things like bi-directional linking. Search, ironically, is also a bit of a mess.

Google Docs was built as a MS Word competitor and that's what it does best. I love Google Docs and I use it every day, but this is one thing I wouldn't use it for.

We're going to see a lot of this over the next 1-2 years.

Software Engineers suddenly feel like they're fighting for their lives for employment, and time won't be "wasted" maintaining OSS for free.

We all need to eat.

The Department of Defense is now the Department of War. They've made their goals clear.

You are not in defense contracting. You are in the business of war contracting.

Take from that what you will.

So it turns out Anthropic was gaslighting everyone on twitter about this then? Swearing that nothing had changed and people were imagining the models got worse?

"Tim Cook to become Apple Executive Chairman"

*John Ternus to become Apple CEO*

Talk about burying the lede, lmao.

I'll definitely put this into the "good problem to have" category.

The eclipse photos are absolutely jaw-dropping.

Does performance not matter?

What if your AI uses an O(n) algorithm in a function when an O(log n) implementation exists? The output would still be "correct"