Running gpt-oss-20b, qwen3.5 and rizzo-flow 4b locally on a $249 Jetson
The author says a Jetson Orin Nano runs gpt-oss-20b, qwen3.5 and the Jev local clone rizzo-flow locally, replacing $700 a month in subscriptions with $6 of electricity.
this is pure f*cking treasure. jensen huang sells a $249 box that quietly killed my $700 a month ai bill, and now the only thing i pay for ai is $6 of electricity. what i was paying every month: claude max 20x ............ $200 chatgpt pro ............... $200 supergrok heavy ........... $300 total ..................... $700 what i pay now: jetson orin nano super .... $249 once electricity .............. $6/mo subscriptions ............ $0 here's what runs on nvidia's smallest box, all local: gpt-oss-20b ................ openai's open weights openai's own open-weight model, apache 2.0, runs fully offline. the same company charging $200 a month for chatgpt pro gave this one away for free qwen3.5-35b-a3b ............ 3b active 35 billion parameters but only 3 billion active per token, so it answers fast enough for everyday chat and coding help rizzo-flow 4b .............. 4.4gb the local clone of typesafe's jev. yes / no / pick one / score with a probability on every answer, so my agents ask the jetson instead of paying an api fast-browser-use ........... qwen3.5-9b a browser agent that clicks through real websites with zero cloud calls and plugs into claude code and codex as a skill ollama + oriveo ............ my phone one app on my phone pointed straight at the jetson. feels like the chatgpt app, except the server is sitting next to my router claude code, chatgpt pro and supergrok heavy, all three cancelled. one $249 board from jensen, $6 a month in power, and my ai bill went from $700 to $6 models and repos in the comments
This post quotes another post: x.com/starmexxx/status/2108159970017488953