Fully Offline Version

[cyrus104]

I am using this for a homelab but this could also apply for any work location that has proprietary data. I would like to be able to grab the AIO version and have it include llama.cpp baked in.

I setup the machine offline to have docker with the nvidia-runtime. I downloaded some models and want to put them in manually into the model folder.

Right now I get a bunch of errors with the backend and model galleries and the AIO doesn't have a backend in it.

[mudler]

for airgap it's sufficient to carry over the backends directory of a running installation, or optionally prepare the backend directory with the CLI local-ai backends install. These can be carried over, just run LocalAI with the corresponding backend folder

[cyrus104]

OK, thank you so do. As a future enhancement it would be nice to not have to go this part and same with the galleries.

…

On Fri, Apr 3, 2026 at 2:40 AM Ettore Di Giacinto ***@***.***> wrote: *mudler* left a comment (mudler/LocalAI#9201) <#9201 (comment)> for airgap it's sufficient to carry over the backends directory of a running installation, or optionally prepare the backend directory with the CLI local-ai backends install. These can be carried over, just run LocalAI with the corresponding backend folder — Reply to this email directly, view it on GitHub <#9201?email_source=notifications&email_token=AALWQ46DU66NZAL4NDC5EZL4T5MHDA5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTIMJYGIYTMMJRGUZ2M4TFMFZW63VGMF2XI2DPOKSWK5TFNZ2KYZTPN52GK4S7MNWGSY3L#issuecomment-4182161153>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AALWQ46ZDDCS57373QHCSKL4T5MHDAVCNFSM6AAAAACXJYITVGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHM2DCOBSGE3DCMJVGM> . You are receiving this because you authored the thread.Message ID: ***@***.***>