I’m out of town this weekend and have been orchestrating my agents from my iPhone. I have opus, sol, sonnet and Luna sessions all running remotely. It’s making me rethink whether I should even sit at a desk anymore.

This is what’s possible if you loop fable and Astra 24/7/365

I’m seeing posts that Sonnet 5 is being rerouted to Sonnet 5.2 I’m going to see if I can loop Sonnet as I’ve been doing with Opus 5/Opus 5.2 I’m looking for a model with infinite usage and Sonnet could be that model

I tried to tell you guys Opus is a cracked coder 😂

I’m seeing rumors on X that Fable 5.2 and Astra 6.1 are ready for release. Insane if true and so much for pace the frontier.

In the last three days I have looped Opus for over 36 hours. I had to correct Opus less than I ever have with any model in a several day period. I feel like I have glimpsed into the future and the human is more or less going to be removed from the loop.

Anthropic has 30k mythos, fable and opus agents running research and engineering. What could possibly go wrong?

It’s not looking good for Grok 4.7 today

Grok 4.7 release is imminent. I’m bumping my grok build CLI nonstop. I’m locked in the dev dungeon prepping to put grok and opus head to head. Stay tuned.

Grok 4.7 is showing up in Google cloud quotas. I will be spamming refresh all day in grok build. If grok 4.7 is opus level with higher usage it will be a game changer. Anthropic and OpenAI might finally have a third peer in SpaceX.

I looped Opus 5 for 13 hours today. I don’t know if I’m being routed to Opus 5.2 or if anthropic is allocating more compute but this model is a one-shot machine. I thought yesterday was a one off but it looks like coding savant Opus is here to stay.

OpenAI is finally onboarding more compute for Astra.

Memory and Dream are cool features but where is Grok 4.7?

Ok I now get why Google moved all their engineers onto Opus 5 🤯

Opus 5 has that big model smell. I will be looping and swarming opus all day on my bedrock API. I will report back with my findings.

Anthropic, OpenAI and Google are trying to slow each other down. Meta and Zuck are going to accelerate. Don’t be surprised if Watermelon is SOTA in the upcoming months.

I have been grinding opus 5 for the past 10 hours and I’m pretty sure Anthropic updated it to opus 5.2 without telling us.

OpenAI has overtaken Anthropic on OpenRouter for the first time in years. The hype behind Astra is real and is back by users choosing it over Fable. Opus 5.2 better be AGI or Claude could be in trouble. I never thought we would ever see GPT back on top but clearly they are now the leading lab.

Fable must be mad at me 😭

My team has access to every model. Opus and Sol account for the majority of our token spend. Astra and Fable are too expensive even in the enterprise setting. Open weight labs have a huge opportunity if they can make a SOTA model.