Journalism begins where hype ends

,,

The rise of powerful AI will be either the best, or the worst thing, ever to happen to humanity. We do not yet know which."

—Stephen Hawking

France’s OVHcloud Kicks Off Its First LLM Family with Olala mini 7A1B; Launches 55 New Cloud and AI Products

OVHcloud’s AI plans extend from developing specialised language models to expanding GPU infrastructure, cybersecurity and enterprise tools aimed at moving AI projects from prototypes into production.
OVHcloud AI Summit stage displaying the Olala branding, promoting its first family of large language models and 55 new cloud and AI products, as speakers address an audience.
November 21, 2026 04:21 PM IST | Written by Neelam Sharma | Edited by Pratima O Pareek

European cloud computing and hosting provider OVHcloud has announced its first family of large language models (LLMs), Olala, with Olala mini 7A1B as its first model, alongside the launch of 55 new cloud and AI products spanning infrastructure, orchestration, inference and AI-powered platform-as-a-service (PaaS) offerings.

The France-based company presented Olala mini 7A1B and its 55 new cloud and AI products at company’s  AI Summit in Paris on November 19.

The compact model is built on a new architecture and hybrid attention mechanism. “It demonstrates that European companies can innovate and train agentic LLMs from scratch,” the company said in a statement.

Olala mini 7A1B has an architecture with 7 billion parameters, of which 1 billion are active parameters spread across 256 experts per layer. The hybrid attention combines Differential-TPA and Mamba-3 in BF16 format. According to the company, only a fraction of the model is used for each request, making it faster and less memory-hungry.

The model runs on a GPU with as little as 8 GB of memory, reaching 160 tokens per second on an RTX A4500. It can also run without a GPU, on a standard processor, at 50 tokens per second.

The model’s context window is currently 32,000 tokens.

The model was trained on three European public supercomputers – Jupiter, MeluXina and Jean Zay – using GH200, A100 and H100 GPUs. The project has been supported by European public authorities with about €7 million in computing hours.

According to the company, Qwant has already adopted it and uses it to generate instant answers to 3 million queries a day. The model is available as open source on Hugging Face. “The future of inference isn’t one giant model that does everything. It’s the right, specialised model for each business need,” the company said.

Version 1.1, expected by the end of the year, will extend the context window to 128,000 tokens and improve its performance in mathematical reasoning.

“With version 2.0, OVHai LLM intends to develop models specialised in three areas,” the company said.

The three areas include agentic AI for businesses. The second area is cloud DevOps, with versions tailored to developers, architects and operations teams.

The third area is cybersecurity, covering both defensive and offensive aspects.

The family is already aiming for more than 100 billion parameters with version 2.0, the company said.

Octave Klaba, CEO of OVHcloud, said, “AI is strategic for OVH. As well as offering AI services to our customers, we must be able to create models for our own internal needs and for those of our customers. We cannot simply rely on the generosity of certain labs to share their LLMs as open source in the years to come.”

He added that know-how in model training has become widely accessible, and the cost of training has been divided by 20 in three years. “It is now possible to create LLMs while making this activity profitable,” Klaba said.

55 New Cloud and AI Products Launch

The company has also launched 55 cloud-computing and AI products. On the cloud foundation side, 38 launches strengthen compute, storage, networking, security, and private cloud. The other 17 cover the entire AI stack, from infrastructure to applications.

NVIDIA B300 and Rubin GPUs are arriving in both Bare Metal and Public Cloud across three continents. With 110 MW of power capacity, OVHcloud’s data centres can accommodate up to 60,000 B300 or 35,000 Rubin GPUs.

For the past three years, OVHcloud has faced a shortage of chips and reserved them for existing customers rather than using them to win new ones, Klaba said at the summit.

Klaba said the company now had no capital spending or capacity constraints and needed more clients to grow annual revenue by more than 10% and reach €2 billion. He also said OVHcloud was targeting six to 12 startup acquisitions a year, Reuters reported.

Klaba said OVHcloud’s servers face hundreds of thousands of cyberattacks every day, prompting the company to develop a cybersecurity-focused AI model due in early 2027, Reuters reported.

New Virtual Private Cloud (VPC)-based network services include a Web Application Firewall (WAF), a Virtual Private Network (VPN), VPC peering, VPC endpoints, an Application Load Balancer (ALB), a Global Load Balancer and private DNS. OVHcloud Connect is now available at 40 points of presence globally.

For AI inference, AI Fabric, built with Blade.sh, enables Python agent sandbox launches. Public Cloud also offers Serverless Containers, Serverless Functions, and Messaging & Events. AI Endpoints has added approximately 100 new models, including voice and image, along with governance tools. AI inference is also expanded to speech transcription and voice generation.

AI Agents hosts, scales, secures and monitors agents connected to enterprise tools through nearly 480 connectors and MCP servers. AI Search allows businesses to explore their documents by combining semantic and lexical search across hundreds of millions of documents per collection.

Application security features include Anti-DDoS Customer Profiles for tailored protection. Cloud Hardware Security Module (HSM) supports encryption key management, while confidential computing keeps data encrypted during processing. Backup Center, Certificate Manager and PKI complete the line-up to protect data, identities and certificates.

According to the company, the AI roadmap is backed by three OVHai labs, dedicated respectively to LLMs, voice, and ontology-enriched document search.

Yaniv Fdida, chief product and technology officer (CPTO) of OVHcloud, said that in Europe, most AI projects never get beyond the prototype stage because moving into production means juggling several products, several providers, pricing grids and data transfers.

“With a single provider that masters the entire cloud and artificial intelligence chain, we offer an integrated user experience with a European offering that is open, reversible and predictable,” Fdida said.

European governments and companies increasingly scrutinise where AI services are operated and how their data is handled, creating opportunities for regional providers seeking to compete with US cloud and AI firms. The new services will be available on the state-certified cloud, OVHcloud said.

OVHcloud also announced partnerships with cloud-computing company Shadow, Qwant and European Search Perspective (EUSP) that connect the web index, models, infrastructure and use cases and deliver end-to-end sovereign AI, fully hosted and operated in Europe.

Also Read: From OpenAI, Microsoft to Global Media Leaders: Paris AI Forum Tackles How AI is Rewriting Rules of News

Authors

  • Neelam Sharma, reporter at AI FrontPage

    Neelam Sharma is a passionate storyteller, and journalist with over a decade of experience across leading Indian media houses.
    Known for her calm presence on screen and powerful storytelling off it, Neelam brings a rare blend of credibility, creativity, and empathy to journalism. Her strength lies in ground reporting and research-driven narratives that connect with the heart of the audience. Whether covering social issues, human-interest features, or breaking news, she combines factual depth with a human touch—making every story not just informative.

    LinkedIn

  • Pratima Pareek, Editor and Co-founder of AI FrontPage

    Pratima O Pareek is an Editor and Co-Founder of AI FrontPage. A gold medalist in Mass Communication and Journalism, she's worked across national and international newsrooms, bringing sharp editorial instincts and a commitment to clarity. She believes in cutting through the noise to deliver stories that actually matter.
    Off the clock, she watches offbeat cinema, follows tennis, and explores new places like a traveler, not a tourist.

    LinkedIn