OVHcloud releases Olala, its first open-weight AI model
OVHcloud has released Olala mini 7A1B, a 7-billion-parameter model under the Apache 2.0 licence, with its weights available on Hugging Face.
According to the press, the model activates about one billion parameters per request and runs on a graphics processor with 8 GB of memory. Qwant uses it for its AI summaries, with answers 20% faster and costs divided by three, figures supplied by the partners themselves. No benchmark score was published at the announcement, made at OVHcloud's AI Summit on 19 November. OVHcloud also plans coding tools based on the model.
Source: Next, published on November 19, 2026.