New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
OVH launches European LLM: Olala mini 7A1B
averagedatahoarder
Member
in News
https://x.com/olesovhcom/status/2122573582911869173
Zero benchmarks, lol. Disappointing
Comments
Yeah, leading with architecture specs and tokens-per-second claims without a single quality eval doesn't tell us much about how useful it actually is yet.
On paper, the footprint is at least interesting for low-end / self-hosted setups — 7B total / 1B active MoE (Differential-TPA + Mamba-3 hybrid), 32k context, Apache 2.0, and targeting 8 GB VRAM or ~50 tok/s on CPU. If it doesn't fall apart on basic reasoning once independent evals hit Hugging Face, a fast 1B-active open model could be handy for cheap boxes.
Still, naming it "Olala" and shipping zero benchmarks in the launch post is peak OVH energy.
bro shut the fuck up
I mean, honestly, were you excited for real about OVH doing an LLM model? I think everyone but themselves expected it to be trash
any english version? i don't speak arabic unfortunately