HumanOstrich

Thanks. Looks like the current Gemini Nano is actually a separate model with the Gemma 3n architecture that has been distilled from Gemini 2.5 Flash[1].

Also, the next version of Gemini Nano will be based directly on Gemma 4 (so not distilled, not Gemini at all except for the name)[2].

So no, it's not a frontier model. Those don't run on your phone or in your browser.

[1]: https://developer.android.com/blog/posts/ml-kit-s-prompt-api...

[2]: https://android-developers.googleblog.com/2026/04/AI-Core-De...

Sources for your claim that the model being downloaded to Android/Chrome is Gemini instead of Gemma. Other than downloading the bin file myself and analyzing it lol.

Can you provide any sources for that? I'd like to learn more about this open frontier model.

Great!

Using that logic, Node.js is just a wrapper over V8.

It's not my plan.

Things aren't just "good" or "bad". There are tradeoffs to consider.

This is the kind of argument you should take to your 3rd grade teacher.

Thanks for all the logical fallacies in one comment.

They don't show your usage in tokens for Claude Code and Codex subscriptions, but that is how they are doing the accounting.

As of yesterday subagents were often getting the entire session copied to them. Happened to me when 2 turns with Claude spawned a subagent, caused 2 compactions, and burned 15% of my 5-hour limit (Max 5x).

They still might pull an Anthropic move and send a C&D or DMCA to archive.org.

Nix is a nightmare and so it usually gets paired with Rust for people who have a complexity fetish.

The article you linked says the change is rolling out in April in evaluation mode.

And if it were related to some kind of scan and malware flagging, the cert would have been revoked. It is not.

From TFA: "I have encountered some challenges but the most serious one is that Microsoft terminated the account I have used for years to sign Windows drivers and the bootloader."

Are you having a stroke?

I wonder if running the compaction in a degraded state produces a subpar summary to continue with.

Did the receipt show it as being a gift? There's a lot of fraud happening the past few months with Claude Code Gift purchases. Anthropic support is ignoring all of it and just not responding to support requests.

Happened to a close friend of mine. A bit of digging revealed the same pattern with fraudulent gift purchases for several other people before I stopped looking. They were also being ignored by Anthropic support. One since January.

Apparently they're so short on inference resources they can't run their support bots. Maybe banning usage of Claude Code with Claude will allow them to catch up on those gift fraud tickets.

Took a long time for me to reach this level of scathing. It is not unwarranted.

Yes, that's why we are angry. Stop making excuses for them.

Well now you're just being silly and I can't take you seriously.

The only "black box" here is Anthropic. At least an LLM's performance and consistency can be established by statistical methods.

After wondering what the heck glitch-freedom is and learning about it, I agree with you. It seems like it deserves at least a brief explanation in an article about how signals work.

I've gone with the universal `alien-signals` package for my project (which doesn't use a frontend framework that includes signals). They show benchmarks of being by far the fastest and have strict limits on code complexity. Those limits are also supposed to avoid glitches by design, and now at least some of that is tested[1].

[1]: https://github.com/stackblitz/alien-signals/pull/39

You no listen. Me give up. Go learn on fruit phone.

> most examples involve the person doing weird things

That doesn't make any sense. There are still plenty of people who are not doing weird things and getting screwed over.

Google fanboys have always fallen for their "don't be evil" nonsense, and it looks like they still are even after it was changed.

Fair point. We only have clear evidence they're being more transparent about credit pricing and value, but it's unclear whether that'll make people burn through usage faster or slower.

The fuzziness is intentional. It gives them wiggle room and obscures how much "value" you actually get from $200, a 5-hour block, or a week. That keeps the tension manageable between subscription pricing and pay-per-token API pricing, especially for larger businesses on enterprise plans who want transparent $-per-MTok rates.

If they were fully transparent, like "your $200 sub gets you up to $2,000 of equivalent API usage," it would be a constant fight. People would track pennies and scream any time 5-hour blocks got throttled during peak hours. Businesses would push harder for pay-per-token discounts seeing that juicy $200 sub value.

1. Look at the new rate card for how many credits are used for each category (that's what the discussion is about).

2. Look at some of your typical sessions for token counts and calculate how many credits that would have been.

3. Look at the rates for extra credits (that's the only place credits have a price).

4. See that you are getting more than $200/mo worth of credits where we have evidence of the value of a credit.

If that doesn't clear it up, then I can't help, sorry.

> effectively $20/$200 in credits for codex

That's not true.

First of all, there's no dollar amount tied to how many credits you get for a subscription.

Second, if you look at the prices for bundles of _extra_ credits and then do some math on the Codex rate card, you'll see that there's no way they would work out to be the same or similar.

It's not a price hike, it's actually making it easier to understand relative usage for different models/features.

I agree with your sentiment, but I wanted to call out that they've always been just as evil as other big tech companies.

I think their motto of "don't be evil" was some pretty clever PR.

I started questioning it c. 2008 when they ghosted me on resolving an issue with my blogspot site that was a bug in the platform. All I could get was a condescending non-response from a "diamond" volunteer on a forum. They were apparently the gatekeepers to reaching actual support.

Yea you don't know anything about LLM architectures. They often change with each model release.

You also aren't aware that there's more to it than "LLM architecture". And you're rather confident despite your lack of knowledge.

You're like the old LLMs before ChatGPT was released that were kinda neat, but usually wrong and overconfident about it.