Jan AI

USER signal comes almost entirely from Jan's ProductHunt launch thread (19 of 20 comments) plus 2 YouTube comments, so it is dominated by launch-day goodwill — roughly half the comments are pure congratulations ('amazing idea congratulations on your launch', 'Looks pretty cool') with zero usage detail. The substantive reports: (1) Setup & usability — 'Easy to set up and use on Ubuntu', 'Works flawlessly, simplest local AI I've installed so far', and the top comment (+6) praises 'no steep learning curves'. (2) Hardware/performance — one user reports 16GB of RAM is enough to run Mistral 7B at ~200 words per minute locally, 'obviously not as fast as a cloud provider'; another hit trouble with the tinyllama model ('something is amiss') before switching to Mistral; a third asks 'What specs do you need to run this?' — indicating spec uncertainty among prospects. (3) Architecture & project health — an informed comment describes the two-part design (Jan UI + 'Nitro' extension for loading local models), and another user checked the repo and found 30+ contributors and high activity. (4) Features — with an OpenAI API key, 'GPT4 works fine out of the box'; the OpenAI integration is cited as a key plus; but image upload and document support are explicitly reported missing; a mobile version is only speculated about ('could be a real game changer'). (5) Philosophy/business — '100% free & open source, no paywalls' is repeatedly cited as the core appeal, with users openly asking how the business model works. (6) Negatives — the only strongly negative datapoint is a YouTube comment: 'Jan is complete crap....I broke it in 5 minutes.' Net read: positive-leaning but shallow; heavy launch bias; the credible hands-on reports (install, RAM, wpm, missing features) are consistent with a promising but early-stage local-LLM client.

No aggregate ratings were found for this product during the last harvest.