
The people around you quietly shape your standards. Choose carefully.

Founder, AutonoMax Innovations. Building @CoRoamApp, AI agents, and useful software. Tech, fitness, philosophy.

The people around you quietly shape your standards. Choose carefully.

Woah… these Astra 6.2 benchmarks are pretty nutty. Looks like it pretty much demolishes Astra 6 and for significantly less.
GPT-6.2 Astra—a frontier model built for your hardest projects. It sets new highs on evals across professional work and scientific research at less than a third of the cost per task of GPT‑6 Astra. You’ll see the difference in things like slideshows, websites, and 3D designs.

Do the frontier labs want hardware prices to go up so it becomes harder and harder to run local AI?

Anyone got a spare dgx spark cluster I can borrow?

Would love to run this locally, just need the hardware to do it…
Qwen4-Max outperforms Qwen3.8-Max by +7.5% net on Agent Arena, making Alibaba #5 lab overall.

Anthropic has been killing it recently. Heading into their IPO, the timing couldn’t be any better Opus 5.5 and Fable 5.2 solo’s everyone and everything… Which lab are you most bearish on right now?

Love our animal friends 😎
the scenic route comes with occasional company.

Just when we thought Terra was being retired…
GPT-6.1 Terra: delivering fast and capable computer use at a lower price. GPT-6.1 Terra understands what’s on screen, takes action across apps, writes naturally, and is a more capable model for your whole workday with the efficiency to take on more of it.

One of my favorite things about @CoRoamApp is building my atlas. Seeing where I’ve roamed and picking new spots I want to explore was something worth building 🌎 🚶♂️

I’m blown away

I need convincing that this is not AI…
BREAKING: Tesla's Roadster goes airborne, soaring over a row of Tesla vehicles in a live stunt at its unveiling event.

For a wall-mounted home screen, voice has to do most of the work. If I'm standing in the hallway typing on a tiny keyboard, I'll just use my phone.
HomeView (wall-mounted) in action.

Retries keep the same photo ID and check Storage + the database before writing again. A lost response doesn't mean you need a new upload.The demo uses a simulated cloud so you don't need credentials. The repo also has a real Supabase adapter and failure tests.

Built a small @expo + @supabase starter for when a photo uploads but the app thinks it failed. It keeps the photo on your phone and checks what's already saved before retrying, even after a restart. Code's on GitHub if you want to try it: github.com/maxgorriaran/expo…

Build a body you're proud of and a bulletproof mind to power it. The goal is to feel capable of taking on anything, and build the strength and discipline to back that belief up.

Microsoft’s approach here is interesting: combine models from different labs with specialists like Decision-1, then put them to work across software businesses already use. Curious how far that goes inside Copilot.
Introducing Microsoft-Decision-1, our new model for fast decision-making. It delivers top performance on structured decision tasks, outperforming both LLMs and other decision models in latency and quality. We’re already testing it across Microsoft for everything from incident response and quality control to scientific discovery.