Win up to €1,000,000 in GPU hours

H company launched a couple of agents in the past few months:

• Surfer H to navigate the web
• Runner H to move from running instructions to performing sequence of actions
• Tester H to verify the results of automated tasks on corporate website.

What is behind Surfer H? Holo1, an Open Source action vLM designed for deep web UI understanding and precise localization. Shared on Hugging Face in June 2025, this family of models has been trained on Scaleway Sovereign GPU infrastructure.

Moshi, Kyutai’s revolutionary AI voice assistant brings unprecedented vocal capabilities. Trained using Scaleway’s high-performance Cluster and served with L4 GPU instances, Moshi excels in conveying emotions and accents with 300x codec compression. This setup enabled Moshi to process 70 different emotions and accents with ultra-low latency, allowing for seamless, human-like conversations. Thanks to this high-performance environment, Kyutai was able to achieve this breakthrough.

Mistral AI, used Nabu, a Custom-built Cluster, to build its Mixtral model (Mistral — Mixtral 8x7B., Jiang et al., 2024), a highly efficient mixture-of-experts model. At its release, Mixtral outperformed existing closed and open weight models across most benchmarks, offering superior performance with fewer active parameters, making it a major innovation in the field of AI. The collaboration with Scaleway enabled Mistral to scale its training efficiently, allowing Mixtral to achieve groundbreaking results in record time.