nickandbro

Very impressive! Interesting how all other benchmarks it seems to surpass Opus 4.7 except SWE-Bench Pro (Public). You would think that doing so well at Cyber, it would naturally possess more abilities there. Wonder what makes up the actual difference there

I asked it to make me a xkcd comic:

https://chatgpt.com/s/m_69e8cc31dac48191a09bb9c00d5aa3fe

kinda funny, I guess

Reminds me of when hedge funds started laying increasingly shorter fiber-optic cable lines to achieve the lowest possible latency for high-frequency trading.

I am curious what workloads Citadel Securities is running on these TPUs? Are you telling me they need the latest TPUs for market insights?

A lot of people are saying it’s disconnected, but even if it was, if a string of your country’s top rocket experts started disappearing, you wouldn’t just sit idly by

200+ points in Arena.ai , that's incredible. They are cleaning house with this model

Thanks for clarifying, so I guess I already have it on then.

Thanks for the feedback, I’ll try to get that sorted out.

I didn't use a game engine

Love this, I will now use backface culling for my game:

https://slitherworld.com

Whenever I read your articles, I get distracted by the space invaders and just play that instead. Maybe this is a problem with me being a bit ADHD, but I feel like I am not the only one

Wow, if the benchmarks checkout with the vibes, this could almost be like a Deepseek moment with Chinese AI now being neck and neck with SOTA US lab made models

Love Hetzner. Cheapest prices in all the land (aside from Hosting your own server) from what I’ve gathered online. Host:

https://slitherworld.com

My foray into multiplayer games.

Here you go folks:

https://www.svgviewer.dev/s/odDIA7FR

"create a svg of a pelican riding on a bicycle" - Opus 4.7 (adaptive thinking)

Very cool, though wouldn't using durable objects for a MMO type game become prohibitively expensive vs using websockets with a stateful server? I assume your game is not sending that many requests so its not too bad.

Pieter levels would love this.

Dang man, chill.

Their website is whack:

https://seguritech.com/en

I want it

Love vim stye editing

Good thing they didn’t bring copilot with them

Yeah working on a smart way to rate limit stale requests for those who don't have accounts.But the final version will allow anybody who is not a bot, to get into a vim instance without logging in. Thanks for the feedback.

If anybody wants to checkout my site to learn the basics of vim. Here it is:

https://vimgolf.ai

I proxy to neovim instances for each level. Still working out some kinks but soon to complete it

I feel like we are just inching closer and closer to a world where rapid iteration of software will be by default. Like for example a trusted user makes feedback -> feedback gets curated into a ticket by an AI agent, then turned into a PR by an Agent, then reviewed by an Agent, before being deployed by an Agent. We are maybe one or two steps from the flywheel being completed. Or maybe we are already there.

Without getting into some of the other things mentioned in the article,

I don't think Vim is going away. Even with all the AI code written, engineers navigate through Claude Code / Codex using Vim (ex: Vim mode in Claude Code).

I really like Vim so much that I've built a gamified way to learn it at https://vimgolf.ai that I am working on completing.

This looks amazing man, seriously good job with this.

very cool

First thing is going outside. Staying inside is only going to compound depression and cause days gone by to become a blur. Just doing that is a step in the right direction, bonus points if it involves moving the body.

Second, start writing. Writing can help prevent circular thoughts and force the brain to plan how to change your lifestyle and get you to a life worth living. If you don’t write, it’s very easy to repeat patterns over and over again.

Third, don’t lose hope. Having a positive outlook and a growth mindset is going to help you navigate setback and push through obstacles. It’s easy to resort to a fixed mindset, but the saints are the sinners who keep on trying as they say.

Best of luck!

yep, just double checked used gpt-5.4 xhigh. Though had to select it in codex as don't have access to it on the chatgpt app or web version yet. It's possible that whatever code harness codex uses, messed with it.