Grok Bot is so good that I wish I could use it on WhatsApp too. It's literally been helping me organize my life in ways I never even imagined were possible.

Day 18: - Decisions API is now generally available - OpenResponses has joined the Linux Foundation - Codex got performance improvements - Uber is going all-in on Codex Another usage reset lands tomorrow before 10 AM Pacific Time (2 PM in Brazil), together with a "small ship".

Anthropic tried to warn the world about just how dangerous they are, and that they were about to unleash a cyber apocalypse.

Anthropic really thought they could quietly slash our usage limits just because we've been hyping up Fable 5.5. Looks like all that success has gone to their heads.

OPUS 5.5'S USAGE LIMITS ARE GETTING WORSE. I'm increasingly convinced Anthropic is running A/B tests with new usage limits, and we're the lucky guinea pigs watching our quotas disappear faster than ever. The team says nothing has changed. Then why are so many power users suddenly burning through their limits like crazy? Several of my colleagues have actually tested how many tokens they were getting on their plans before versus now, and the difference is huge.

We're getting Kimi-K3.1-Opus-5.5-GPT-6-Astra any day now, guys. (I've actually tested Kimi K3.1, and the model is genuinely INCREDIBLE...)

Incredible news for OpenAI! This massive investment means more capacity to serve users, better usage limits across plans, new GPT models, faster inference, and even more products and features. A while ago, I mentioned that more compute would be coming online. That's already happened, and there's a lot more on the way.

I no longer believe Anthropic is secretly routing Fable 5.5 through Haiku, Sonnet, and Opus in Claude Code. I think they're just rolling out newer checkpoints with more recent knowledge cutoffs, running continuous A/B tests, and improving their models every day. Asking a model what Token Gremlin posted last week and getting an answer doesn't magically mean you're talking to Fable. At this point, we're all going crazy waiting for a model that's taking so long to launch it's practically becoming a legend.

I'll stop hyping up Fable 5.5 the moment OpenAI releases a new model that beats Opus 5.5.

Translation: Fable 5.5 is apparently such an absolute MONSTER that it can finish in HOURS what would take human experts YEARS, check its own work, fix its mistakes, and hand you almost bug-free code. And Anthropic somehow made this sound like just another Thursday update.

In other words, Fable 5.5 is going to be so ridiculously smart that you won't need to babysit Claude anymore. Claude will babysit YOU, do all the work, and occasionally ask if you're sure you should be allowed near a computer.

Today marks exactly one month since Opus 5.5 was released to the public, and still nobody has managed to dethrone it. WE NEED A REAL CHALLENGER. We can't just let Anthropic rule the AI world for this long.

Four interesting updates: - "Kibitz" (rumored to be Kimi K3.1) is an absolute BEAST at everything related to design. It reminds me a lot of Opus 5.5 and GPT-6 Astra at the same time. (Distillation?) - The Gemini 4 Carbon checkpoint is better than Argon, but honestly, it's not all that impressive compared to Opus 5.5. - "Kalmia" (rumored to be Qwen 4) is absolutely insane when it comes to design, Three.js, and similar tasks. - The checkpoint believed to be Fable 5.5 also seems much better than before, but it's nowhere near as far ahead of Opus 5.5 as people might expect. In other words, Chinese models have broken through the Fable 5 barrier. They're now just as good, while being ASTRONOMICALLY cheaper. As for Fable 5.5, I think its biggest challenge won't be beating the GPTs. It'll be outperforming its predecessor, Opus 5.5, by such a huge margin that it actually justifies the price. And honestly, I don't think it'll pull that off.