Model Terminal

Robostral Navigate

Robostral Navigate is Mistral AI's first embodied navigation model — an 8B vision-language-action system that takes a single RGB camera image and a plain-language instruction to drive a robot through indoor or outdoor environments without LiDAR, depth sensors, or a prebuilt map. Built entirely in-house, it targets autonomous navigation in offices, warehouses, and outdoor spaces. Source: supabase.

Value score
Not scored
Context
tokens
Max output
tokens
Price

Capability radar

Identity

Developer
Mistral AI
Openness
Modalities
Release
Knowledge cutoff
Deprecation
API docs

Benchmarks

No benchmark scores yet.

Compare nearby

Pricing

Input / 1M
Output / 1M
Speed