Is the Lenovo Thinkcentre Tiny M75Q Gen 2 good for a first functional homelab (with also a small local AI model)?
retained text
Hi everyone, I want to build a home server for years, and the more I discover about this world, the more it gets confusing when talking about what to do and to pick. I personally love these tiny form factor, and the modularity, plus the great design put into this little guy made me wanna aim for one fo these, and this m75q gen 2 seems to be the most balanced between price and performances (the Gen 5 looks hotter, but it also costs basically double the price 😅), more than the chinese mini pcs that looks cool and all, but I don't know how are them in terms of affidability over a 24/7 active working, but maybe I'm just biased, you will tell me ahah Ok, my ideal budget is around 300-500€ (the 500€ stretch is for the mandatory upgrades I would need, like the ram... I'm already sweating), and my needs will be pi-hole+tailscale, media stuff like stremio, immich and jellyfin, a personal cloud/nas/thing to have those comfortable 1-2tb of safe and private storage, if possible some streaming gaming (but this is clearly what these minipc are not for, so I won't cry if it'd be better ditching this), and if I'm lucky with the ram aspect, a little local ai that could assist me during work (graphic designer that typesets books, so mostly I'd need it to sort files and folders, checking docs and pdfs for a round of typos, wrong accents and similar things, so no generative crap), and mahbe do some vibecoding personal projects when needed. Feel free to suggest me other cool services I surely don't know! Regarding the AI model here I need the most help, I understood some basis about Qwen, the 1 to 4 bit stuff and the need of 128gb of ram for unleashing it's true power, but I still don't know what it actually can do, everyone talking about this stuff is always so vague or clearly talking to an audience that already digested everything about the topic. Is it possible to install a good model that could work well for my needs, without having to get 64 or 128gb of memory, especially in such lenovo tiny? Would the other services in the server be castrated by the presence of the model using the rest of the free ram? Would the presence of a local model force me to get a newer mini pc? Man I feel so out of touch regarding AI 😅 Thabks in advance for any clarification guys, I'm in your hands 🙏
document fetched 2026-10-28 00:36 · canonical https://reddit.com/r/homelab/comments/1xgpk76/is_the_lenovo_thinkcentre_tiny_m75q_gen_2_good