Artificial intelligence / news brief
OVHcloud Announces Its First Open Model, Olala Mini
- Event
- Published

OVHcloud announced Olala mini 7A1B, a mixture-of-experts language model with 7 billion parameters and 1 billion active per request. The company says it runs on an 8GB GPU, reaches 160 tokens per second on an RTX A4500 and is already used by Qwant. Weights, a technical report and AI Act documentation are promised for November 19, while no quality benchmarks are available yet.
Why it matters
A major European cloud provider is moving into model development, but the promised open release and performance claims cannot be assessed until the artifacts arrive.
Sources
- OVH announces Olala mini 7A1B (external site)OVHcloud · primary source ·
- OVHcloud launches European-trained open source AI model (external site)ITPro · coverage ·
- OVH unveils Olala, its first open source AI model (external site)Numerama · coverage ·
- OVHcloud unveils Olala mini, its first LLM: speeds, no scores (external site)ActuIA · coverage ·