OVH launches Olala mini 7A1B: open-source European 7B LLM that runs on CPU
jedisct1 · x · 2026-11-19
OVH's OVHai LLM has released Olala mini 7A1B, the first model in a new family built from scratch in France and trained on three EuroHPC supercomputers (Jupiter, MeluXina, Jean Zay), positioning itself as sovereign European AI.
- Five models in the family: generalist mini, Base, SFT checkpoint, plus 18-language translation and structured summarization variants fine-tuned from mini
- Frugal architecture: 7B params with sparse activation per request — runs on GPUs with as little as 8GB RAM (e.g. RTX A4500) or even CPU-only, targeting on-prem, laptop, and edge deployment
- Open source: available on Hugging Face with a technical report; fine-tuning on custom data offered
Related event: OVHcloud Releases Open-Source European Model Olala mini(3 posts)→
More from Infra
- One Amazon EBS gp3 command scales a 10M-vector Weaviate index 26×, from 9.8 to 260 QPS — philipvollet · 2026-11-19
- Ben Goertzel investigates why FAA, Microsoft, Apple, Chase and more all crashed in the same week — bengoertzel · 2026-11-19
- Researcher begs for spare compute as university cluster stalls, ~18 GPU-hours left to finish — vxnuaj · 2026-11-19
- Local MiniMax H3 on 6GB VRAM: 50 minutes tuned vs 6.5 hours on defaults — No_Cow26 · 2026-11-19
- Compress Wikipedia to ~27GB locally: neural compression via tufazip — Terminator857 · 2026-11-19
- One GPU transcribes 8.7 hours of mock GP calls in 18 minutes, 27.8x realtime — MaziyarPanahi · 2026-11-19