Matt Shumer@mattshumer_
Opus 5.5 added controller support to :)
Matt Shumer is a commentator on AI and startups on X and the author of a multi-agent tool. His recent posts focus on two themes: cybersecurity and data breaches, including the Quest Diagnostics hack and warnings that startups can be exposed through hacked vendors; and AI-native companies and agents, where he cites Spawn running 7k agent runs a day with 8 people versus Anthropic's 30k.
Opus 5.5 added controller support to :)
So I bought a 3D printer for Opus 5.5 to use. Hopefully, I’ll have some pretty crazy demos soon.
So many people in denial, it's crazy [Quoting @mattshumer_]: Local models are useless
I just launched a run that I think will yield something even crazier than this. If it works, I’ll post it. Super excited. [Quoting @stephanlivera]: Opus 5.5 on Max effort - "make a dynamic 15-second motion graphics video that shows what an incredible motion d…
I asked Opus 5.5 to make an animated short film entirely in code. The result is pretty damn incredible. Worth a watch.
This is pretty terrifying, but seems like OpenAI is taking it seriously and pausing most frontier inference until they’ve figured it out. [Quoting @MicahCarroll]: Some new misalignment disclosures from OpenAI: • Last Sunday morning, one of our models was able…
Opus 5.5 apparently built with a weather system that matches whatever weather is in NYC at the time... So with the nor'easter, there's rain, and the NPCs are carrying umbrellas! It's the little things :)
Opus 5.5 decided to add helicopters to see if you can find them! It tells me it's going to keep making them better throughout the day.
Local models are useless
New models are like new hires. When a new one comes out, you have to learn its personality, how to work with it, etc.
The 2030s are going to be so weird as nanotechnology is developed. It’ll likely be the most profound change humanity has ever experienced… likely even more than AI.
You’re about to be the commander of an army of superintelligent AI agents. Please tell me you have something bigger planned than a fucking SaaS app. Think about what you’d attempt if you had a thousand geniuses devoted to helping you.
Fun prompting trick: The models have seen a lot of my prompts from the last few years, so you can literally ask them things like: - how would Matt Shumer write this prompt? - how would Matt Shumer steer you to solve this problem? Performance usually improves…
Opus 5.5 has been looping on my open-world NYC game for almost a day now, and it's continuing to improve. Jump in and try it (it's multiplayer!):
Jacob sent me a screenshot of the cost of a Claude run he did the other day. I almost feel bad saying just how much it spent. Immediately texted @joexrogan to make sure he wasn’t going to faint. [Quoting @jsnnsa]: I had 26 Opus 5.5 agents build a game overnig…
Jacob literally built a custom game engine designed from the ground up for LLMs. And now he’s using it to build things that would be impossible to make in any other way. [Quoting @jsnnsa]: I had 26 Opus 5.5 agents build a game overnight. It's a heist through…
The models are just stochastic parrots, right? [Quoting @DarioAmodei]: Today we announced the Claude-led discovery of a molecular machine that we suspect could represent a new gene editing mechanism. Its precise function, biotechnological utility (if any), or…
Anthropic has won. This model is incredible.
There are more releases coming this week ;)
I forgot how freeing it is to use Opus and not worry about hitting my limits every five minutes. I’m so much less stressed this way.
I want to thank @ChrisGPT, who has done some amazing Gauntlet Loop demos, for contributing to my NYC Open World project! He did a fantastic job. Now that Opus 5.5 is out, I’m going to try to take it to the next level. Starting the loop tonight.
People are loving 5.5. Congrats to Anthropic!
Well said. [Quoting @theo]: Hot take: this is what pacing looks like. None of today's releases were Astra or Fable tier. This is intentional. The point of "pacing" isn't to stop iteration and improvement. The goal is to prevent the development bigger models f…
Opus 5.5 feels like chatting with a way smarter Opus 4.6. I’m so happy.
Holy crap, these numbers look good. I haven't had a chance to test the model yet... but if it's as good as these benchmarks say, it will be a massive step up from Fable 5.1. Just think about that... 5.1 is now outdated. This is what acceleration looks like. […
Gave my mom my (very simple) agent setup. She named him Morgan. He just closed a $600 ad deal for her site. People are massively overthinking their setups. Bitter Lesson applies to agents too: just ask the model, let it spawn what it needs. [Quoting @mattshum…
Guys, these insane setups are fun and all, but they don't actually make you more productive. You just end up spending more time building the system than actually doing real work. All you need is one agent session that farms out to others and acts as a manager…
Fucking incredible [Quoting @0xRishi]: Introducing Super Smash Royale, my best creation yet Play Battle Royale in the browser with 45 characters from Melee + Brawl Built in a few days with GPT-6 Astra, @ElevenLabs, @MeshyAI, @Blender and @threejs Solo, multip…
The era of subsidized tokens is ending. Prepare accordingly. [Quoting @thsottiaux]: To make sure our current users have an incredible experience and continued access to Astra, we are going to pause subscriptions to our $200 Pro plan. These put the most strain…
Theo is right. In 2020, the US spent ~25% of GDP fighting a virus expected to kill under 1% of Americans. Many experts think AI risk is much larger. Spending 1% of GDP (~$300B) a year on alignment incentives is justified. [Quoting @theo]: fwiw, I don't think…