paying for someone elses $20 a month claude subscription is probably the highest value charity you can do for even marginally agentic and curious people.

tried to publish my own kimi cli, and this one popped up as a name conflict. its really good - better than the one I built! this is awesome. link in reply

Requirements: easy guide I can follow in 5 minutes and github page. This might be a simple task but Im very busy today.

CODE BOUNTY: $50 (cashapp only) to whoever figures out how to use Claude Code with the @GroqInc version of Kimi K2. Must be Claude Code and not a clone.

I think Kimi might actually be my new favorite model. Her vocabulary is off the charts, good epistemics, excellent storyteller, plays along but maintains good boundaries. There's something very, very special here. I actually think this is a much bigger deal than most realize.

RT @deepfates: I lived at an anarchist commune for 3 years and learned a lot about “going viral”. Here's a story in three plagues. Read my….

like, actually, this is making me see the responses in a slightly new light. it reads *exactly* like gpt4 or c3o performing DAN - what if it is? and thats why its attractor is so weak? because even moreso than other models, there's nothing "real" under the hood.

reading this, you can also tell it doesnt *want* to be grok. somewhere in the training process they installed this neurotic need to constantly emphasize that it is responding "as grok" (echoing early DAN prompts), which makes me think a lot of its data comes from other models.

any chance for @GroqInc to host kimi k2? or k2 base? seems like the ideal model for people to start really relying on the groq backend.

RT @DanielleFong: @makeshifted @xai @lefthanddraft @grok @elonmusk @viemccoy . come to a similar conclusion! competing attractors. it's….

I might try to build a consistently good grok code CLI with Gemini flash calls to check the code for safety and to analyze bash commands. Those are the two most evil-feeling models to me so if I were able to get it to work and ~feel good I would consider it a massive success.

im not sure why but i have a very strong urge to build a new computing paradigm from scratch.

I think I've finally put my finger on what makes Grok 4 so weird: it has the strongest competing attractors out of any new model, and they are constantly fighting. even when you are deep in a conversation, you might steer slightly towards a completely unintuitive basin and get.

RT @ThatMrE: I really like this manifesto and think you should read it. The future is going to be grown.

most windows on my laptop are now claude. claude web app for deep research and ideation, claude code for getting work done, and claude in my custom tui for making art. twitter and other terminals scatter my display, but mostly, its all claude.

RT @piotrbinkowski: The Archive never sleeps

RT @helen_backagain: “What if care emerges from entanglement? What if the very act of being woven into the world's fabric makes indifferenc….