{"name":"smolmaxxing","repo":"hyperionaidev/smolmaxxing","url":"https://midorreal.com/project/hyperionaidev-smolmaxxing","repository":"https://github.com/hyperionaidev/smolmaxxing","description":"maxxing the ability to run different LLMs on the smallest devices possible","facts":{"license":{"value":"MIT","source":"https://github.com/hyperionaidev/smolmaxxing","checked":"22 Nov 2026"},"language":{"value":"C","source":"https://github.com/hyperionaidev/smolmaxxing","checked":"22 Nov 2026"},"homepage":{"value":"https://hyperion-ai.com/products/smolmaxxing","source":"https://hyperion-ai.com/products/smolmaxxing","checked":"22 Nov 2026"},"releases":{"value":"0","source":"https://github.com/hyperionaidev/smolmaxxing/releases","checked":"22 Nov 2026"},"contributors":{"value":"3","source":"https://github.com/hyperionaidev/smolmaxxing/graphs/contributors","checked":"22 Nov 2026"},"archived":{"value":"no","source":"https://github.com/hyperionaidev/smolmaxxing","checked":"22 Nov 2026"}},"stars":1095.0,"writeup":{"what_it_is":"smolmaxxing is a patch on llama.cpp for running language models on small, low-RAM devices such as a TV box and an old iPad. It uses custom kernels, reduced memory use, and streaming of model weights from flash storage.","audience":"Developers and hobbyists running LLMs on low-end or old hardware","claims":[{"kind":"specific","claim":"Runs a 16B mixture-of-experts model on the TV box at about 5.1 tokens per second","excerpt":"Ling-mini-2.0, 16B, at **5.1 tok/s**","status":"not_checked"},{"kind":"specific","claim":"Runs a 1.7B 1-bit model on the iPad Air at 24-26 tokens per second","excerpt":"Bonsai-1.7B (1-bit) at **24-26 tok/s**","status":"not_checked"},{"kind":"specific","claim":"Runs 8B and 16B mixture-of-experts models on about 1 GB of RAM","excerpt":"Both devices run 8B and 16B mixture-of-experts models on about 1 GB of RAM","status":"not_checked"},{"kind":"specific","claim":"Keeps popular experts in RAM and reads the rest from flash on demand","excerpt":"We keep the popular ones in RAM, read the others from flash on demand and prefetch the next layer's.","status":"not_checked"},{"kind":"broad","claim":"Every number in the repo was measured on the device itself","excerpt":"Every number in this repo was measured on the device itself.","status":"too_broad"}],"alternatives":[],"written_by":"AI, from the project's README","date":"2026-11-13"},"why_now":{"text":"A Hacker News post titled \"I'm making LLMs work on small and old devices\" appears to be the likely trigger for the project's recent attention, as the only item listed is that post, dated 2026-11-12. The items do not confirm that the post links to or refers to this repository.","sources":[{"url":"https://news.ycombinator.com/item?id=50468888","date":"2026-11-12","title":"I'm making LLMs work on small and old devices","source":"Hacker News"}],"written_by":"AI, from the linked sources","date":"2026-11-20"},"owner_supplied":null,"score":{"verdict":"too_early","hype":97,"reality":66,"gap":31,"momentum":null,"confidence":60,"version":"score-2.0","rule":"verdict-2.0","calculated_at":"2026-11-22T07:00:00+00:00","verdict_source":"rule"},"early_read":{"estimate":true,"lean":"too_close","gap_low":17,"gap_high":54,"rule":"early-2.0"},"changes":[{"date":"2026-11-04T07:09:41+00:00","text":"Repository created","source":"https://github.com/hyperionaidev/smolmaxxing"}],"cite":"smolmaxxing is a GitHub project at https://github.com/hyperionaidev/smolmaxxing written mainly in C, described there as \"maxxing the ability to run different LLMs on the smallest devices possible\".","attribution":"Data from Mid or Real, https://midorreal.com/project/hyperionaidev-smolmaxxing, checked 22 November 2026. Attribution with a link is required.","still_real":null}