Bryan Bryce – e/tacos

@nikitabier @dieworkwear @grok how to view Country of Origin for a user

When I first started at X, this is one of the things I did a deep dive on because I felt it was critical to the integrity of the experienced. I tasked our Threat Disruption team to investigate a number of trends that seemed artificial. The findings: We could not find meaningful examples of foreign interference in US policy discussions, except people gaming rev share in developing countries. This is what motivated the release of the Country of Origin feature and significant changes to the rev share algorithm. The most deranged & divisive replies generally were from residential IPs in the United States—with no signs of using a VPN. Ultimately, X is a reflection of the internet. And that means you will see the full spectrum of human thought. And sometimes the most outrageous takes will catch fire. Having said all of this, there can still be cases of narratives being boosted but the origin of the initial post is almost always domestic and we have hardened our systems in the last 3 months to prevent this.

the one conspiracy i believe is that this place is full of bots programmed to say insane, incendiary things and they're funded by foreign organizations who want to divide americans and destabilize US society

I solved the model picker problem, no need to thank me Tibo

@maria_rcks @thsottiaux

@haider1 No, it isn’t, you haven’t used it

GPT-5.6 Luna is a standout model for price-to-performance at half the cost, "5.6 luna max" matches gpt-5.5 xhigh, while outperforming "fable 5 md" at half the cost -- so together, "luna" and "sol" dominate the Pareto frontier i'm still unsure when to use "terra"

@mylifcc No

Gpt-5.6-Luna max的能力和Sol-high一样高，但费用是1/3，都高于5.5 如果不是特别高的要求，使用luna可以获得3倍的总量

In the conversation about whether or not or to what degree AI will result in some engineers losing their jobs, most people seem take a fairly naive angle. They focus on model intelligence and autonomy as a comparison to human intelligence. But just "making software engineering 10% faster and 10% more expensive" is enough to affect the equation. Companies don't have infinite demand for software 🧐 It's a practical certainty that some jobs will scale down/up as a result of the AI adoption, regardless of what stance you take. If AI is a hyped up slop machine then there will plenty of work to be had cleaning up it's mess. If it's the singularity then there will be plenty of work directing the token firehose, or you'll go into business for yourself and run your own token firehose 🤷‍♂️ A dramatic simplification but the point is to expect widespread and varied disruption to the software industry over time, no matter how the cards fall.

@james_bohrman @thinkymachines Ur fun, insta follow

@BryanJBryce @thinkymachines You didn't read the paper did you? Again, not everyone is a tourist dad of 7 who works with tech as a hobby.

The model works in more real situations too, like when Mianna is trying to say words she's only read before.

@james_bohrman @thinkymachines Talk to us, oh wait, whahaha why is it people who use the word larp are the ones larping? What did you study in college poly sci?

@james_bohrman @thinkymachines lol, can’t reach the top bar p and y are clipped, cope harder

@james_bohrman @thinkymachines How am I supposed to enter my email in that tiny box, is it a form for ants?

@james_bohrman @thinkymachines People use mobile bro

@BryanJBryce @thinkymachines Oh why am I not surprised. Vibe coded slop site: brycelabs.com

@aarondfrancis Nah, it’s still what to build because it requires extensive interactions with humans and human attention, always was

I don't want to sound crazy but I'm starting to think that writing code maybe was in fact the bottleneck

@james_bohrman @thinkymachines I did research on the model and realized that it’s HUGE, it’s laughable that you would use this model

@BryanJBryce @thinkymachines Why are you asking me questions right after trying to tell me what the reality is? Do you realize how stupid that is? And yes, that's exactly what I'm saying. Have you ever actually implemented YOLO and an LLM? It's the dumbest implementation for this.

@james_bohrman @thinkymachines You couldn’t even run this model on consumer hardware, you can run YOLO on a Jetson Nano 😆😂😆😂

@BryanJBryce @thinkymachines My brother in delusion YOLO is like a decade old algorithm and no it can not. Do you even know anything about YOLO? Can you tell me the slightest thing about the latency tradeoffs of using a architecture that's meant for text for something like this? No, because you're a larp.

@james_bohrman @thinkymachines Ur a moron, YOLO latency is tens of milliseconds, the video is sampled by this model every 200 ms 🤣

@cjzafir Horrible take, same tasks take 2x longer on Luna xhigh than Sol med

A correction after robust testing. Replace GPT 5.6 Terra (High) with GPT 5.6 Luna (Extra High) for execution. Same performance but: > 1.3x faster than Terra > 2.5x cheaper than Terra In Codex, you burn 2.5x less limits if you use GPT 5.6 Luna (Extra High) as subagents.

@james_bohrman @thinkymachines Literally what they’re showing in that video is possible with yolo and an LLM ¯\_(ツ)_/¯

@BryanJBryce @thinkymachines No lol read the paper and try actually building something

@jrxcket Does it run on coal?

The Chinese method of catching boosters using the wires is actually kinda cool lmao