Scaleway expands Managed Inference with “Bring Your Own Model”

Scaleway has launched a new feature for its Managed Inference platform: Bring Your Own Model (BYOM), giving developers and enterprises even more flexibility when deploying AI workloads in a secure and sovereign environment.

Managed Inference is Scaleway’s dedicated offer for private AI endpoints. Each deployment is one model, on one GPU, for one client. This approach guarantees performance, privacy, and full control over deployment.

Until now, users could deploy leading open-weight models like Llama, Gemma, Mistral, Deepseek, or Qwen. With the introduction of BYOM, they can now deploy any compatible model directly from Hugging Face, the most comprehensive catalog of AI models in the world.

Hugging Face has played a key role in making AI open and accessible. As a strong supporter of open source, Scaleway shares this vision and is committed to empowering builders across Europe with transparent, interoperable, and developer-friendly tools.

The new BYOM feature unlocks powerful use cases for enterprises, from fine-tuned models to custom RAG pipelines and multimodal agents. Each model runs in isolation, on dedicated infrastructure, with no rate limits and a predictable price.

“Bring Your Own Model is about giving builders the freedom to innovate on their own terms,” explained Franck Pagny, Product Manager for Inference at Scaleway. “By integrating with Hugging Face, we’re making it easier than ever for teams to deploy the models they want, within the infrastructure they trust.”

Try it now and bring your own model to life.