@hnykda this isn't getting enough coverage. Tools like OpenClaw and Hermes use LiteLLM and people give these tools their digital lives

I genuinely don't get the same "vibe" that some l33t ppl our there are getting where GPT-5.4 is superior to Opus 4.6. I've been using both back & forth either coding, code-reviewing, planning, designing/specc'ing . starts strong feels great, 90k tokens in & then petters out

made my computer dramatically play BBC news music before every meeting

this is pretty brutal, maybe this is why their LLMs are the best.. I think this adversarial approach they take on AI-safety makes them stronger.

built a local dictation tool for linux. whisper.cpp on vulkan + a Go binary. press a key, talk, text stream into the focused window. no cloud, ~2.5s latency. vibe-coded. https://t.co/0TW7zKX4et

We need widespread usage of this "SWE-CI: Evaluating Agent Capabilities in Maintaining Codebases via Continuous Integration" https://t.co/xDIVwSAlQ9 But in more broadly, having a more longer-term evaluation of the impact of Agent-based software development is URGENT NEED.

Portugal’s New AI Tool Flagged €110 Billion in Suspicious Contracts https://t.co/8L2PIqIP2H

I think one of the funniest things is how frontier labs spout how AI is writing their code, but the reality is that the products they ship are buggy, resource hungry messes. It's kind of the worst advertisement ever for both their products and their worldview.

Isto e' escandaloso! Como e' possivel este tipo de incompetencia. Estes relatorios sao de uma importancia extrema para garantir que acidentes raros (mas de alto impacto!) nao voltem a acontecer. https://t.co/A1jLqWKnzQ

llmfit. Useful tool that probes hardware and tells you exactly which LLMs will actually run. - handles MoE expert offloading, picks the best quantization for your RAM, estimates tokens/sec before you even pull the weights. Essential for local dev. https://t.co/qn8PmqQNbV

This 𝗖𝗟𝗔𝗨𝗗𝗘.𝗺𝗱 file will make you 10x engineer 👇 It combines all the best practices shared by Claude Code creator: Boris Cherny (creator of Claude Code at Anthropic) shared on X internal best practices and workflows he and his team actually use with Claude Code daily.

And Part two... How and Why Local LLMs perform on a framework13 strix point laptop.. I learned a lot on exploring this and trying to make things go faster... https://t.co/6ppRjiDOe8

#Benchmarking Local LLMs for coding in Go on my framework13 AMD Strix Point laptop... #framework #linux #golang https://t.co/RvLXPPvVjd

always fun and entertaining .. besides obviously informative :-)

Thinking that a subscription to the best AI models for coding turns you into a professional programming machine… is like thinking that a subscription to Mathematica 20 years ago turned you into a professional mathematician. Good luck trying to solve problems you don’t

I also practice daily serializing and deserializing a Binary Search Tree on paper. I love it!

For people tell us that “well, Chomsky has not been proven wrong, and his model of language is still valid”… Let us examine Chomsky’s New York Times article… The False Promise of ChatGPT (2023). Almost the entire essay is rambling pseudo-philosophy where Chomsky gives us vague

🇦🇷Buenos Aires delivered. #DuneCon25 set the pace for what onchain data can be. The Onchain Data Platform, in its purest form. Thank you for being part of it.

If the main scaling constraint for AI is power, what does this chart tell us?

This is nice joke.. quite the reference to that LKML email..