OVHcloud's OVHai open-sources Olala-v1.0-mini-7A1B: hybrid Mamba-3 MoE reasoning model with ~1B active params
Skyline34rGt · reddit · 2026-11-19
OVHcloud's OVHai LLM released Olala-v1.0-mini-7A1B, an open-weights (Apache 2.0) small reasoning model with GGUF builds available.
Key points:
- Architecture: hybrid Mamba-3 / attention MoE; only 7 of 36 layers keep a KV cache; 1B active params per token, runs on a laptop
- Training: designed in France, trained on three European supercomputers (JUPITER, Jean Zay, MeluXina) on 12.8T tokens of public and synthetic data
- Capabilities: chat, instruction following, and tool use; thinks before answering with four reasoning-effort levels; RL-trained for multi-turn tool calling and search
Related event: OVHcloud Releases Open-Source European Model Olala mini(3 posts)→
More from Models
- Gemini 4 breakdown: Neon is 2x faster than Carbon but hallucinates more — teortaxesTex · 2026-11-19
- Gemini 4 Neon (High) beats Hydrogen and Carbon on AI SEO advice in hands-on test — gaganghotra_ · 2026-11-19
- Agents spent years hillclimbmaxxing while the human at the keyboard became an afterthought — tokenbender · 2026-11-19
- Speculation: Argon/Carbon/Neon may be nested distillations, with 'adapted' dodging 'post-trained' — teortaxesTex · 2026-11-19
- 6.2-Astra reportedly crushes contest benchmarks, finding a novel AWTF26 solution in 2 hours — i_dg23 · 2026-11-19
- KVARK Flash: a proxy that caps DeepSeek's thinking per request, 30-50% fewer tokens — Ill_Contribution_717 · 2026-11-19