technocrat8080

Wow, would love to see a source for this.

Have they ever talked about their size or weights?

5.4 Mini's OSWorld score is a pleasant surprise. When SOTA scores were still ~30-40 models were too slow and inaccurate for realtime computer use agents (rip Operator/Agent). Curious if anyone's been using these in production.

You can provide the screencapture cli as a tool to Claude and it will take screenshots (of specific windows) to verify things visually.

In my experience, especially with Opus 4.6, using subagents greatly mitigates the startup context hit. 4.6 has very obviously been RL'ed on subagent usage and it almost always spins up an Explore agent to get a feel of the codebase and get a token-efficient summary. The 1M context version of 4.6 further alleviates this.

My original question was more along the lines of implementing things like PR review yourself. I was tinkering with an internal service that spins up ephemeral CC instances to analyze PRs, but realized this can easily generalize across arbitrary tasks. Was curious what sort of things folks could use that for.

To be clear, it's not a TEE replacement but does address one of the most common use cases of TEEs

Seems pretty obvious Sky.app's functionality will land in the macOS ChatGPT app at some point. I wonder how Atlas fits into that story.

Sky is macOS only. It essentially gives an LLM access to various system APIs coupled with a floating user interface that you can access on command.

By parallel, do you mean editing the codebase in parallel? Does it use some kind of mechanism to prevent collisions (e.g. work trees)?

What kind of agents do you have setup?

Curious about this too – does it use the standard context management tools that ship with Claude Code? At 200K context size (or 1M for the beta version), I'm really interested in the techniques used to run it for 30 hours.

A bit confused, all this to say you folks use standard containerization?

I sense the undertone here is that there's a cost to genetic testing – did I get that right? Could you expand on that philosophy?

My god, how can I take photos like that? Do you describe your setup and process somewhere?

I mostly agree with your first point, but is there not a chance for more serious disease to be detected? In which case specialized preventative care is the best path forward? There is definitely something to be said about staying ignorant of these, however.

> The only people who should not have kids are bad parents and people who don't want kids.

I don't think people who are highly susceptible to birthing malformed children should have children either. Genetic testing helps figure out if you might be in that group.

Thanks for spurring the thought. The main reason I'd like to find out is to avoid bringing a child into the world that might suffer from incurable disease.

Largely agree with your point on preventative actions being the same with or without testing, but there is a tail end that might warrant specialized action.

I only care about testing for genetic health issues. I don't want my DNA to be inserted in a database that can be cross-referenced to check for relatives, ancestry, etc.

How do you folks think about the Manus finding on dynamic tool selection? https://manus.im/blog/Context-Engineering-for-AI-Agents-Less...

> A natural reaction is to design a dynamic action space—perhaps loading tools on demand using something RAG-like. We tried that in Manus too. But our experiments suggest a clear rule: unless absolutely necessary, avoid dynamically adding or removing tools mid-iteration. There are two main reasons for this:

> 1. In most LLMs, tool definitions live near the front of the context after serialization, typically before or after the system prompt. So any change will invalidate the KV-cache for all subsequent actions and observations.

> 2. When previous actions and observations still refer to tools that are no longer defined in the current context, the model gets confused. Without constrained decoding, this often leads to schema violations or hallucinated actions.

> To solve this while still improving action selection, Manus uses a context-aware state machine to manage tool availability. Rather than removing tools, it masks the token logits during decoding to prevent (or enforce) the selection of certain actions based on the current context.

A 100k fee is well within the territory of killing job prospects for skilled foreign students graduating from US universities.

What percentage of the AI labs are staffed by either foreign workers or second/third generation immigrants? Look at the composition of high achieving high school students- almost certainly of Asian or Indian descent, certainly many belonging to families of recent immigrants. The pipeline this EO disrupts is immense.

Great, let's punish early stage startups instead of rewarding the successful ones.

How are startups supposed to afford this? How are talented H-1B workers supposed to start companies? And no, the answer is not always an O-1. I know plenty of foreign founders contributing meaningfully to the US economy, now slapped with a 100k fee.

Why can't the phone itself act as the input device? Sure it's not a full-fledged keyboard, but it could work wonderfully in a pinch.

Along with the ParetoQ paper from Meta (https://arxiv.org/abs/2502.02631), the concept of low-bit LLMs seems to be gaining traction. Has anyone experimented with this in production? I'm aware of a few pre-transformer era companies focused on applying this to CNNs

Vastly overstating its capabilities? SF is ~crawling~ with them 24/7 and I've yet to meet someone who's had a bad experience in one of them. They operate more than well enough to replace rideshare drivers, and they have been.

There goes building somewhat featureful toy kernels for your RPi