OVH releases Olala-v1.0-mini-7A1B, an open hybrid Mamba-3 MoE reasoning model
EveYogaTech · reddit · 2026-11-19
OVH's OVHai LLM released Olala-v1.0-mini-7A1B on Hugging Face, the first of a new model family: a 7B-param small reasoning model for chat, instruction following and tool use, built on a hybrid Mamba-3/attention mixture-of-experts architecture. Designed in France, trained on 12.8T tokens across three European supercomputers (JUPITER, Jean Zay, MeluXina), released under Apache 2.0 with support for 37 languages — a notable European sovereign-AI release.
Related event: OVHcloud Releases Open-Source European Model Olala mini(3 posts)→
More from Models
- FoodTruck Bench, an LLM agent business-sim benchmark, lands on Papers with Code — NielsRogge · 2026-11-19
- GPT's Ultrafast mode is 'really good' but burns through tokens, says dev — lxfater · 2026-11-19
- Same 10k-Token Budget: Hermes Adapter Solves 51% of TB-2.1 vs Base Model's 28% with xhigh Thinking — yb2698 · 2026-11-19
- Hermes adapter solves a SWE-Bench Pro task in 35 tool calls where the base Qwen3.8 model failed — yb2698 · 2026-11-19
- NousResearch's Hermes adapter lifts Qwen3.8-27B agent scores: SWE-Bench Pro 29.4%→43.1% at 4× fewer tokens — yb2698 · 2026-11-19
- Gemini 4 Neon hailed as 'most tasteful model ever' for its essay on thriving in the AI age — daniel_mac8 · 2026-11-19