Paris-based Mistral AI, which sells AI models and computing infrastructure to businesses, said on July 8 that it launched Robostral Navigate, an 8 billion-parameter embodied navigation model for robots.

Mistral said the model works across wheeled, legged, and flying robots.

It uses a “navigation via pointing” method to predict targets or fallback movement commands.

It achieved 79.4% success on seen environments and 76.6% on unseen environments in the Room-to-Room Continuous Environment (R2R-CE) benchmark after training on about 400,000 simulated trajectories across 6,000 scenes.

In May, Mistral said it struck deals with Airbus SE and BMW AG focused on industrial engineering and simulation-heavy research and development, rather than factory-floor robotics.

The work includes technical documentation, aerospace design tools, edge object recognition, and crash-simulation models.

In June, Mistral was reportedly in talks to raise about 3 billion euros (US$3.43 billion) at a 20 billion euros (US$22.9 billion) valuation, after a September funding round valued the company at 11.7 billion euros (US$13.4 billion).

🔗 Source: Bloomberg

Recent Mistral AI developments