bel8

I too love and use pi agent.

But all of these agents have or had "dumb" security issues:

https://github.com/anthropics/claude-code/security/advisorie...

You probably know that since you run pi inside https://github.com/containers/bubblewrap.

The RCE they mention was fixed in December, 2025:

https://github.com/anomalyco/opencode/releases?page=21#relea...

I would agree if openrouter was the only provider of those.

But those LLMs are also offered outside openrouter.

https://openrouter.ai/rankings#top-models

Chinese models dwarf USA models usage. And now there's a Fable/5.6 alternative. The gap widens.

Now go and ask for your GP poster for their data as well. Unless you're only interested in data that supports your bias ofc.

Continued how? I have switched most of my personal LLM coding to DeepSeek V4 Flash since it was launched.

And now 100% to a mix of K3 / DeepSeek V4 / MiMo 2.5.

It's nice not being called a terrorist just because I told it to reverse engineer something.

At work they are still hemorrhaging money to Western providers due to enterprise contracts but I foresee they won't renew for much longer. Specially of the upcoming final version of DeepSeek V4 proves to be Opus+ level.

That's quite small for my workloads.

I try to keep it under 200k but my DeepSeek and MiMo sessions can sometimes grow to 350k tokens when I try to squeeze one last iteration I compact.

Can't OpenAI copy DeepSeek K/V cache tech (from published papers) to make it super cheap?

Contrary to many, I have high hopes that Jarred and team will lapidate the ported codebase and it will continue to flourish as an open source project.

Cool, and how is it ranked?

I used K2.7 to reverse engineer a protocol from Wireshark packages. No complaints. K3 should be fine too.

What harness? These can make or break benchmarks because of tool call failures/limitations.

And is the benchmark open source?

Mozilla exists because Google gives them billions to keep Google as default search engine.

Apple has a decade of beef with EU.

OpenAI has a lot ot catchup on the EU hate scale.

I would respect Apple's privacy talk if they facilitated running Linux. But it takes years of reverse engineering to get it barely useful on a Mac.

So they put revenue above true privacy, as companies do. But they are the loudest to spit the bs while selling black boxes.

I don't see how that invalidates their point.

Besides, n=1 benchmark seems like more of a coin toss.

I guess this is why Anthropic keeps extending their Fable credit window for subscriptions.

They probably knew a Fable contender was coming and hit the panic button, twice.

Apple must be happy that they let Jony Ive go. What a letdown.

(assuming this meh partnership rebranding had his participation)

Yep. That's my case.

I now have unlimited Anthropic and OpenAI plans at work because CEOs bought the hype.

But for personal projects $10/mo OpenCode Go serves me with DeepSeek V4 Flash, MiMo 2.5 and GLM 5.2.

Obligatory mention of Handy which is open-source and cross-platform:

https://github.com/cjpais/Handy

It's great for me when configured to use Parakeet v3.

Rust Bun has been running Claude for some time now.

I'd be inclined to trust production code in such a large usebase.

I disagree. It's fine to contextualise that an opinion solidifies over time.

And your parent is right to point out that emotionally loading posts tend to be a disservice to the message being carried.

I'm left with a good impression.

Lot's of comments on the whys and historic context.

Backed by tons of tests.

Yes the file could use some modularisation and splitting into smaller parts but maybe the gains were weighted and judged against.

That file is a great example of KISS.

I think Odin's batteries included approach have a chance to achieve escape velocity.

Specially if their community and their BDFL continues to be welcoming and fun to interact with.

Their 1.0 roadmap announcement is cool: https://youtu.be/dLPAqXi9In0

Here's most of the language in a single demo file: https://odin-lang.org/docs/demo/

Yeah I don't understand these myopic takes.

Jarred's post about Bun-Zig-Rust post was technical and polite.

Andrew's post in response was anything but that.

They also migrated and that also made the workflow cheaper.

It has everything to do with the article.

I get 95% or more cache hit rate with pi and DeepSeek or MiMo so it doesn't invalidate.

But I'll investigate how that works in a session. You got me curious.

It's easy to add using plugins.

What do you miss? I ask because I do some heavy work with pi + GLM 5.2 (using opencode Go subscription) and my workflow is plan -> implement.

And pi agent is even less.

The entire agent system prompt can be seen here:

https://github.com/earendil-works/pi/blob/main/packages%2Fco...

You're the one downplaying "breaking patents", not me.

I don't understand the correlation.

And by "Figure out without issue" you realize it got the wrong answer right?

As also stated in the article.