{"name":"backburner","repo":"StayLameBro/backburner","url":"https://midorreal.com/project/staylamebro-backburner","repository":"https://github.com/StayLameBro/backburner","description":"Your iPhone helps your Mac run a 27B model: faster prompt reading and more context over a USB-C cable","facts":{"license":{"value":"MIT","source":"https://github.com/StayLameBro/backburner","checked":"22 Nov 2026"},"language":{"value":"Python","source":"https://github.com/StayLameBro/backburner","checked":"22 Nov 2026"},"last_release":{"value":"v0.0.4, 3 Oct 2026","source":"https://github.com/StayLameBro/backburner/releases","checked":"22 Nov 2026"},"releases":{"value":"4","source":"https://github.com/StayLameBro/backburner/releases","checked":"21 Nov 2026"},"contributors":{"value":"4","source":"https://github.com/StayLameBro/backburner/graphs/contributors","checked":"21 Nov 2026"},"archived":{"value":"no","source":"https://github.com/StayLameBro/backburner","checked":"22 Nov 2026"}},"stars":1196.0,"writeup":{"what_it_is":"Backburner is a project that lets an iPhone connected to a Mac over USB-C help run the Qwen3.8-27B model locally. The iPhone runs part of the model's layers during prompt reading and holds older context for attention computation.","audience":"Mac and iPhone users running large local language models","claims":[{"kind":"specific","claim":"Prefill is faster with the iPhone, by 29-44% at 16k-48k context","excerpt":"29-44% faster prefill at 16k-48k","status":"not_checked"},{"kind":"specific","claim":"A 24 GB Mac with an iPhone can hold 196k-229k tokens of 8-bit context, sized from the phone's free memory","excerpt":"196k-229k tokens at 8-bit on an iPhone 17 Pro Max","status":"not_checked"},{"kind":"specific","claim":"Greedy output is token-identical with and without the phone","excerpt":"Greedy output is token-identical with and without the phone","status":"not_checked"},{"kind":"specific","claim":"Supports a range of Mac and iPhone hardware combinations, with some reported by other users","excerpt":"Others\nhave run it on 8 and 18 GB Macs, an M5 and an M1 iPad Pro and an iPhone 18 Pro Max","status":"not_checked"},{"kind":"specific","claim":"Splits model layers between the Mac GPU and the iPhone GPU in a pipelined way","excerpt":"the Mac runs layers 1-40 and the iPhone runs 41-64\non its GPU, pipelined","status":"not_checked"}],"alternatives":[],"written_by":"AI, from the project's README","date":"2026-10-26"},"why_now":null,"owner_supplied":null,"score":{"verdict":"mid","hype":23,"reality":42,"gap":-19,"momentum":0.2,"confidence":60,"version":"score-2.0","rule":"verdict-2.0","calculated_at":"2026-11-22T08:00:00+00:00","verdict_source":"rule"},"early_read":null,"changes":[{"date":"2026-10-03T21:51:20+00:00","text":"Latest release v0.0.4","source":"https://github.com/StayLameBro/backburner/releases"},{"date":"2026-10-03T01:06:56+00:00","text":"First release","source":"https://github.com/StayLameBro/backburner/releases"},{"date":"2026-10-01T20:59:56+00:00","text":"Repository created","source":"https://github.com/StayLameBro/backburner"}],"cite":"backburner is a GitHub project at https://github.com/StayLameBro/backburner written mainly in Python, described there as \"Your iPhone helps your Mac run a 27B model: faster prompt reading and more context over a USB-C cable\".","attribution":"Data from Mid or Real, https://midorreal.com/project/staylamebro-backburner, checked 22 November 2026. Attribution with a link is required.","still_real":null}