Needed a small footprint llm proxy to surf free models for simple tasks at home -> https://t.co/3Kbv6GkchC . tiny executable, rust, fallback chains on rate limits. #llm
iPhone 16 pro max failed doing a ‘restore’ on macOS finder (4046). Now device is undetectable in finder, apple configurator (2 systems), apple devices (win). Multiple cables. Quite shocked it’s fragile, had expected more with official apple software,.,.
People are saying that Nigel Farage has asked Elon Musk to ban people from X for sharing this clip of him attacking his crony Robert Jenrick.
Surely free speech champion Nigel Farage isn't that petty? But just in case, THIS is the video that you mustn't download or share
Obama stock trades as president: 0.
Biden stock trades as president: 0.
Trump stock trades in Q1 2026 alone: 3,600+.
When do we all just admit he is a grifter and half of the American population got tricked into thinking he'd help anyone but himself and his inner circle?
@DivyanshT91162 I’ve been building more tools for myself - as well as using other rust-developed (or at least efficient, compiled) tools from others. the differences in footprint - ram/cpu can be amazing.
@SubhamSaho4836 Hard to know what to think. it changes constantly. old cli aliases don;’t work, quotas are all over the place. Too confusing to invest time in imo
@TheRealJamieKay Lack of freedom. The freedom to live, work, visit a wide range of countries with fascinating cultures. Hope, ideas - all vapourised by brexit
@adrgrondin@glifocat@lmstudio what inference engine did you use? What model format, quantization, what about context length & lv cache parameters? (on a 64GB m4 max best I’ve got is with qwen 3.6 30b/A3b, 128-256k context, 4 bit mlx under omlx)