Today in AI

New to Codex? Here’s 7 tips from shipping apps with it.

Paul Solt@PaulSolt

I use many of these skills when my agents get stuck. If you’re building iOS APPS, install these skills from @twannl and @twostraws

Paul Solt@PaulSolt

tl;dr: Fable 5 basically mocks every other model and tops every benchmark. Curious to see whether the six new benchmarks will crown a new leader after GPT-5.6.

Artificial Analysis@ArtificialAnlys

Introducing six new Artificial Analysis Capability Indices for comparing model capabilities across key industry domains.

Interesting thing I heard from a founder of an early-stage AI startup in SF: Hiring is damn hard for them b/c the type of devs that pass the hiring bar for their founding engineer (onsite) role are... the types of devs who also happen to hired by the AI labs in SF or get accepted to YC as well! And so most folks choose those options instead...

GitHub PRs should have read receipts I want the bots to know when they’ve been left on read

Codex remote supports SSH login keys now! Been waiting for this since Codex mobile was first released

Thomas Ricouard@Dimillian

Big release for Codex Remote in the latest ChatGPT iOS update! We added many things. Threads management tools are now available (hello Chief of Staff). You can now filter your diff with unstaged, staged, branches, etc. Support for SSH keys login, and much, much more!

In addition to GPT-5.6, OpenAI’s new voice model also seems very exciting. It works “bidi,” meaning bidirectionally, which is exactly what we had hoped for from the voice model back then. Of course, I don’t know whether it might be released together with 5.6, but the first use cases and examples featuring the new voice model were extremely promising. I’m very excited about the new model! h/t @testingcatalog

🚨 AI News | TestingCatalog@testingcatalog

OPENAI 🔥: GPT Bidi 1, an upcoming bidirectional voice mode for ChatGPT, has been renamed to GPT Live 1!

How do folks run AI-assisted engineering interviews these days?

Check out a great overview of some of the new, recently shipped features and changes in Hermes Agent!

Alex Finn@AlexFinn

MASSIVE Hermes Agent update over the last few days.

Ding ding ding! After more than a decade of required reviews in various codebases, I think there's very little a human can do better in a review than a GPT-5.6. (It's different when reviews are used for sharing knowledge or discussing approaches.)

Nate Berkopec@nateberkopec

This is the version of the “software factory” that I think is realistic btw. I’m not sure what human review adds to dependabot PRs, for example. We should be able to build a factory that merges these within 5 minutes of them being opened.

This is awesome. Setting up local models can be tricky, so anything that makes it more accessible is a really great thing. @MikeBradleyAI and Ahmad are doing awesome work. Check them out.

Ahmad@TheAhmadOsman

Local AI was never this EASY.

ok I think I have ai pyschosis reason? I think its so fun building a harness to play my game that I am not even making much progress on my game. I think I am a tools dev in my heart and setting up something to run CLI commands / talk to server just touches the fun part of my brain

i figured out exactly what made ralph so effective. you can invert the latent space of the model and it’s shockingly powerful. iykyk

Get more done with Codex MANAGER threads. Use a "chief of staff" thread to manage other work in Codex. Here's how I'm using it to fix up my app, Super Easy Timer, while I sleep. Bookmark this article and try it yourself. Prompt included.

Paul Solt@PaulSolt

Add OCR to your own projects using Apple Silicon. If you're doing anything with TEXT in images, don't waste tokens with OpenAI or Anthropic when you can do it on-device. Check out this awesome text recognition project from Jeffrey (and follow him for next-level agentic loops) @doodlestein

Jeffrey Emanuel@doodlestein

I keep growing the scope of my new FrankenOCR project.