There\u2019s an onboard custom-built capacitor board to store some juice, mind you, but it only provides around 20 seconds of crank-free runtime before you\u2019ve gotta keep crankin\u2019 to keep it alive. That, and it takes a bit of time to get it running – according to the documentation website, it\u2019s a 30-second process \u201cfrom the moment you start cranking to the moment you\u2019re having a conversation with CrankGPT.\u201d\u00a0<\/p>\n
According to the AI expert duo behind the device, computer scientist Katrin Tomanek and former Google Advanced Technology and Projects Group technical project lead Alex Kauffmann, CrankGPT still delivers impressive results despite the need to perform some hard physical labor for your tokens (though we\u2019d argue some exercise for your AI might not be a bad thing).\u00a0<\/p>\n<\/p>\n
\u201cAsking Claude to add two numbers for you is like swatting a fly with a wrecking ball,\u201d Kauffmann told The Register in an email. This tongue-in-cheek demonstration, Kauffmann said, may be a bit of light fun, but it\u2019s an exercise in demonstrating what his and Tomanek\u2019s AI company, Squeez, is all about: small, private specialized AI models that, in a pinch, might not even need very much energy or a connection to the web to operate.\u00a0<\/p>\n
\u201cSqueez produces customized, efficient, and private models that can run on small, inexpensive hardware to solve specific problems,\u201d Kauffmann explained, citing tasks like voice recognition for someone with a strong accent or speech impediment, or specially-trained, local AIs that are subject matter experts in topics like gardening or auto repair, but won\u2019t touch subjects outside their wheelhouse.<\/p>\n
Contrary to the flashy dot-com for CrankGPT<\/a> the pair have set up, Kauffmann told me, Squeez has no plans to pursue spin cycle class-powered AI stacks for dev teams, though he said if anyone wants to foot the bill, he’d be happy to give it a shot.<\/p>\n “Off-the-shelf bike generators are shockingly expensive and they’re fussy to build,” Kauffmann said. Still, “a good biker can maintain a steady 120W output, so a class of twenty could power a Blackwell.”\u00a0<\/p>\n Speaking of wheelhouses, what\u2019s inside that box?\u00a0<\/p>\n If there\u2019s a tiny computer in a 3D-printed box with a crank attached, there\u2019s a good possibility it\u2019s going to be a Raspberry Pi, and that\u2019s the case here. CrankGPT\u2019s brain is built on a stock RPi 5 with 8 GB of RAM and a cooling fan HAT, and audio input and output are handled by a dedicated I\/O HAT<\/a> designed for voice assistants running RPis. Power comes from the aforementioned crank, which is actually an off-the-shelf<\/a> 20W switchable voltage hand crank unit built for emergency USB device charging, and is stored in the custom capacitor unit the duo built.\u00a0<\/p>\n \u201cThe neatest part of the whole thing is that you can actually feel the inference,\u201d Kauffmann told us. \u201cThe amount of resistance the crank presents varies depending on the amount of work the board is doing, so when it’s really working (generating words for instance), the crank becomes much harder to turn than when it’s idling waiting for you to say something.\u201d<\/p>\n \n The guts of the CrankGPT machine As for software, the device is running the most stripped-down, bare bones instance of DietPi the pair could compile, which is able to boot into a functional userspace in about three seconds. The voice agent is the truly original piece of work done for the project, as detailed in the documentation page, and was built entirely from scratch.\u00a0<\/p>\n \u201cWe wanted to understand the system end to end and have as few dependencies as possible,\u201d the documentation page notes. It\u2019s available on GitHub<\/a> for those interested in trying it out.\u00a0<\/p>\n Speech recognition is handled by the Moonshine<\/a> automatic speech recognition engine, chosen for its speed, while text-to-speech synthesis is handled by Piper<\/a>, chosen again for its low-resource edge inference capabilities.\u00a0<\/p>\n As for the models running on the thinking itself, there are a few that are behind CrankGPT, with Liquid LFM2 1.2B providing a general-purpose voice agent, and Gemma 3 1B being used for translation. CrankGPT can switch between translation and various prompts (e.g., general question answering and games like two truths and a lie) via a knob on the side of the enclosure.\u00a0<\/p>\n \u201cIt\u2019s entirely configurable,\u201d Kauffmann told us. \u201cWe added a couple of physical inputs (the knob, a button, a switch) to make experimentation easier.\u201d<\/p>\n Kauffmann added that he and Tomanek were surprised by how well the translation function worked. \u201cWe did no fine tuning, it’s just a two-line prompt and it works really well for high-coverage languages,\u201d he explained.\u00a0<\/p>\n While the demonstration focuses on audio prompts and responses, Kauffmann explained that the device supports all sorts of different models, with the only real limitation being inference time and the amount of hand cranking one wants to do to get their response.\u00a0<\/p>\n \u201cWe\u2019ve generated images (small), made poetry (bad), and written code using the same setup,\u201d the CrankGPT makers wrote in their documentation, all with \u201ca hand crank, a little computer, and a small stack of speech and language models running locally.\u201d\u00a0<\/p>\n If you\u2019re interested in building your own CrankGPT model, keep an eye on the documentation page we linked earlier in this story, as Kauffmann told us he and Tomanek are planning to release all the plans and schematics in the coming days, while the aforementioned custom voice agent is already available for tinkering.\u00a0<\/p>\n \u201cIt’s a pretty straightforward setup, the only tricky part is that SBCs like the Raspberry Pi will sometimes draw enough current to trigger a little generator’s overcurrent protection,\u201d Kauffmann told us. If you have a spare $300 lying around (that\u2019s what Kauffmann estimates the RAM pricing surge has driven the build cost up to, from the $150 he spent when building CrankGPT last year), then you, too, may soon be able to build your own completely off-grid, standalone AI box so you can keep chatting with your favorite micro LLM if and when its bigger cousins knock the grid offline. \u00ae<\/p>\n","protected":false},"excerpt":{"rendered":"Datacenters got you down? Worried that even the most innocuous questions will spin up AI models running in…\n","protected":false},"author":2,"featured_media":634287,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[20],"tags":[554,733,4308,86,56,54,55],"class_list":["post-634286","post","type-post","status-publish","format-standard","has-post-thumbnail","category-artificial-intelligence","tag-ai","tag-artificial-intelligence","tag-artificialintelligence","tag-technology","tag-uk","tag-united-kingdom","tag-unitedkingdom"],"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/posts\/634286","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/comments?post=634286"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/posts\/634286\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/media\/634287"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/media?parent=634286"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/categories?post=634286"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/tags?post=634286"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}![\"A](\"https:\/\/www.newsbeep.com\/uk\/wp-content\/uploads\/2026\/06\/5254344.webp\")
<\/p>\n
\n Source: Squeez\/Katrin Tomanek\/Alex Kauffmann\n <\/p>\n