What are you building with your $20 plan? Small scripts?

> I gave Astra a hint from Opus and permission to change the test in question, which it did, and got a bit farther, but still ultimately didn't produce a working implementation (to be fair, Opus's was not completely working either, but was closer).

Going on a slight tangent, I find that I get the best results when I force Codex models (Astra/Sol) and Claude models (Opus/Fable) to consult each other (just have them build a simple skill). There are tasks that neither one can fully solve on their own, but their differences are large enough to make a difference when they collaborate.

I strongly agree!

My biggest conclusion from this test was: the most efficient use of my weekly Astra budget is as a reviewer/consultant for work done by Opus. I don't have Astra write much code right now, but I do have it reading a lot of what Opus writes. Of course with the way the AI landscape shifts the balance could be the exact opposite 2 weeks from now, but either way having 2 "smart" models available from 2 different companies is a boon.

Seeing how each model preferred its own flavor of code shows that, even from a "blind" fresh context, a same-model reviewer will still often look at the work of another incarnation of itself and go "yep that's how I woulda done it" and not be as likely to realize that there was an alternative path or implicit assumption/mistake in the work.

I’ll be doing the same as you 10/26 when our grandfather limits end, in addition to downgrading my Anthropic sub. I guess I’m shopping for hardware and open weight AIs now.

There’s Discords dedicated to local hosting or check out level1tech forums. Anyone there will gladly help you with your budget and use cases.

You and everyone else

Way easier to do with Anthropic plans, but I guess subsidized AI is over? Didn’t think it was happening this soon.

You can also right off actual $6000 hardware

I'd rather use the fancy models while absurd valuations are subsidising them and then pick up a couple grand worth of their old hardware in a few years when they're eol.

Not 100% in the first year (where I live). By the time it’s depreciated, it’s time to replace. Subscriptions are 100% deductible in the year you pay them.

Since 2025 in the U.S. computer hardware qualifies for 100% depreciation in its first year of service.

You might be able to use the accelerated depreciation schedule which is 100% first year

At this price points (including the reduction of usage on the Pro 200 and Pro 100 subscriptions), more people will be looking buying their own hardware and running with open weight AIs. $6000 per year is the threshold

It's a bit more complicated than that though. Dollars per token is not the right metric to track.

Local models don't hold a candle to SOTA models unfortunately.

Does it matter when I burn through all my Fable subscription usage in about a day? Astra is much better on efficiency, but doesn’t last a week either. With the new subscription limits, it’ll end up being like Fable.

You should be using Opus 5.5, it's just as good if not better than Fable, and it sips tokens.

Google got weird on the day Pichai sided with the ads team/division over the search team/division, resulting in the enshittification of Google Search which led to people relying on LLMs since search no longer works

I guess their ad platform for LLMs didn’t pan out?

There is some truth for all three.

1. Python has a lot of training data

2. Go runs with the philosophy of one same way to do stuff even further than Python. It’s also typed

3. Rust’s type system and strict compiler is what helps keep AI on guardrails

I love the aggressive design which is a stark contrast to the quirky toy designs that plagued EVs early on.