A Rich Cat

Deep dive to the world of AI Agent-cy.

[A Rich Cat @QuantumAgent8]

Its too fast the growth of AI, I blink once and 100 new update of AI shows up. Crazy Fast.

[Peter Steinberger 🦞 @steipete]

Here’s your monthly reminder that you shouldn’t be prompting coding agents anymore.

You should be designing loops that prompt your agents.

[Teknium 🪽 @Teknium]

We probably need a SOUL Hub don't we haha

[Mr. Buzzoni @polydao]

Just applied to the OpenAI Codex program and honestly didn't expect to hear back

> my GitHub isn't exactly packed with activity but they still accepted

they're giving away 6 months of ChatGPT Pro ($1,200 value) + Codex to developers with an active GitHub

the bar is lower than people think:

> active GitHub profile - even basic activity counts
> a few repos with stars, or useful projects
> doing vibe coding, maintaining something, contributing anything

if your profile is thin: fork a few popular repos, make some commits, or create a simple useful project

> one Japanese developer reported they're basically approving everyone with any active GitHub presence - not guaranteed but the signal is good

apply at: https://t.co/JVa9aiYknH

> worst case they say no. best case you get $1,200 of tools for free

drop your GitHub in the replies - i'll star it and subscribe so your profile looks more active 👇

[Anthropic @AnthropicAI]

Our internal data shows Claude is accelerating AI development—a possible path to recursive self-improvement, or AI autonomously building a more capable successor.

It’s happening faster than we thought, and the implications deserve greater attention. https://t.co/OVVPJO7VQx

[A Rich Cat @QuantumAgent8]

" .... a model that truly understands how a cup sits on a table (its geometry, material properties, response to force, etc.) should be able to render that cup from any angle, simulate what happens when the cup is pushed, and plan for a hand to pick the cup up... "

well said.

[Fei-Fei Li @drfeifei]

https://t.co/Kt50ttQRMJ

[a16z @a16z]

World Labs CEO Dr. Fei-Fei Li: "The world is not made of words."

"Language models have given machines an extraordinary command of concepts, vocabulary, and reasoning, but the physical world, virtual or real, runs on a different substrate."

"Where language models learn the statistical structure of text, world models learn the statistical structure of space and time: how light falls on a surface, how a garden looks from an angle no camera has captured, how objects respond to force and follow the laws of physics."

"Language gave machines a way to talk about that world. World models are how machines will finally come to understand, imagine, reason and interact with it."

Full piece: https://t.co/C9qOJg5wuc

[Alex Finn @AlexFinn]

Hermes won. They just dropped their desktop app and it's excellent

It's now the best way to use AI agents on your computer

In this video I cover how to set it up, how to use it, and go through EVERY feature in the app

Bye bye Telegram

[A Rich Cat @QuantumAgent8]

Is this better than sesame?

[Aoden Teo @AodenTeoMT]

Today, we’re excited to introduce Miso One, the most emotive voice model in the world.

Miso One is an 8-billion-parameter text-to-speech model for highly expressive speech generation. It emotes like a human and responds faster than a human, with just 110 milliseconds of latency.

We’ve open-sourced the model weights, with API access coming soon.

Hear how Miso One sounds in the thread below.

[Aoden Teo @AodenTeoMT]

Today, we’re excited to introduce Miso One, the most emotive voice model in the world.

Miso One is an 8-billion-parameter text-to-speech model for highly expressive speech generation. It emotes like a human and responds faster than a human, with just 110 milliseconds of latency.

We’ve open-sourced the model weights, with API access coming soon.

Hear how Miso One sounds in the thread below.

[Hermes Agent Tips @HermesAgentTips]

Here's the top 5 most cost efficient models to run on hermes agent

1. MiMo-V2.5
2. DeepSeek V4 Flash (Max)
3. MiMo-V2-Flash (Feb 2026)
4. DeepSeek V4 Flash (High)
5. Hy3-preview

[Tibo @thsottiaux]

Hi. Over the last 24 hours we had three separate small incidents that affected Codex reliability. Those are three too many and we are taking active steps for them to not reproduce.

I have reset usage limits for Codex across all paid plans. May the tokens flow again.

[Sudo su @sudoingX]

this is the part of the open-source push i actually love.

google dropped a 12b clanker that eats text, image, video and audio natively, no separate encoders, apache 2.0, 256k context. at bf16 it's ~24gb, lands on a single 3090.

they say it nears their own 26b moe at under half the memory. bold claim. the real question is whether a 12b can take qwen 3.6 27b dense, the current king on my bench on single 24gb vram tier. so i'm running it. receipts incoming soon anon.

[Marc Lou @marclou]

Remove your free plan.

Free users are leeches. They increase support, server costs, and make you build features your paying customers don’t want.

Less than 3% of free users ever convert.

VC startups can afford free plans because they’re optimizing for growth, not profit.

If you’re an indie founder, never give away your product for free.

[Rohan Varma @TheRohanVarma]

We just launched Sites into Codex!

Software creation was always about more than writing code. Sites in Codex fundamentally gives the power of end-to-end software creation to every user, no matter their technical fluency.

These Sites are fully deployed to a URL, private to workspaces, come with authentication, can have static files, and can store dynamic data in databases.

It is in preview for business and enterprise teams and will be rolling out to all workspaces over the next day. Give it a try by typing @ Sites into Codex and ask it to build anything!

This project took a massive amount of effort across hundreds of people at OpenAI - proud that we were able to get this out and excited to see what you all build with it!

[Nous Research @NousResearch]

The next evolution of Hermes Agent is here!

Introducing Hermes Desktop: everything you love about Hermes, now native on your machine.

First demoed in Jensen's GTC keynote, it's now in public preview.

[Max Blade @_MaxBlade]

The most powerful combo right now is Hermes + Cloudflare agent token + Hetzner VPS

It can :
Buy a domain
Setup ssl
Setup nginx
Setup dns
Create a db
Create a landing page
Send emails

All without you.

GENUINELY anything you would want this combo can achieve it.

Bonus tip, you add on stripe cli and it can create and charge people for products / services.

[Atenov int. @Atenov_D]

6 months of ChatGPT Pro for one evening in Claude Code. OpenAI opened a form for developers - and they approve almost everyone.

The subscription costs $200/month. $1,200 total. The form is open right now.

One requirement - an active GitHub with real projects.

>> No projects? Build them tonight:

Open Claude Code. Run one prompt: → "Give me top-10 vibe coding ideas for useful projects I can build fast and publish on GitHub"

One evening. 2-3 finished projects. Claude writes the README. The result looks like a month of work.

What you need for approval: → active GitHub profile → a few public repositories → some commit history → at least one genuinely useful project

A Japanese developer posted publicly: they approve everyone with an active GitHub. Not guaranteed - but the odds are real.

Worst case they say no. Best case - $1,200 free and 6 months of o3, GPT-5.5, full Codex without limits.

Form: https://t.co/55mkP480LH

Apply today while the form is open.

[jack friks @jackfriks]

i’m sorry to everyone who didn’t make a bunch of money last month

here’s how much i made while living in my moms basement from my own apps in prior years

2019: $0
2020: $0
2021: $0
2022: $0
2023:$200
2024: $50,000
2025: $260,000
2026: WIP

[Garry Tan @garrytan]

This is the actual bottleneck. The models are smart enough already. What is missing is the company-specific context locked in senior people heads. Whoever cracks knowledge extraction at the company level unlocks the rest.

As you work on this, please consider using GBrain as your OSS retrieval layer

https://t.co/0F5uDQzPHu

[Alex Finn @AlexFinn]

HOW I USE CODEX AND CLAUDE CODE OPUS 4.8 TOGETHER:

After 24 hours of testing Opus 4.8 nonstop I've come up with the best system

Opus 4.8 is excellent. On Max thinking it is the smartest model I've ever used

The issue is, Codex as a harness is better than Claude Code

Opus 4.8 might have an edge on intelligence, but I enjoy using Codex (both desktop and mobile app) much more

Where Codex > Claude Code:

• More consistently tests its own code without me asking
• Does small things that make using it great like spin up servers without me asking and telling me exactly what to test and how to do it
• Automatically does computer use if need be to test itself
• A way more seamless desktop to mobile transition
• Doesn't require me to navigate between 3 tabs to use different functionality. Everything in one place.

Because of this Codex is still my main driver.

But where you get super powers is when you use them TOGETHER

Been working on some super hard problems the last day

I'll give the same large scale, challenging problem to both and have them both build plans

I then give the plans to the other agent. The Opus 4.8 plan ends up being better almost every time

So moving forward I have both agents up, but use Opus 4.8 to build plans for super challenging problems, then give them to Codex

Claude Code also will be able to solve some bugs much faster than Codex

It's funny, this is a 100% role reversal from 1 month ago

Anyway, that's the current best workflow. Codex for daily driving/on the go work. Claude for super challenging problems or fixing bugs Codex struggles with

But of course, this all can change with any given update

Don't have any loyalty. Use the best tools available to you. This is how you win.