Google DeepMind develops Gemini Robotics models for robot action and embodied reasoning. Its portfolio separates vision-language-action control, higher-level planning and on-device inference, so access and evidence must be tracked by exact model revision.
Identity boundary. Added with root approval from the manifest NVIDIA ecosystem substitution queue to represent the verified developer of Gemini Robotics; not an extra speculative company route.
Distinct model roles
The VLA model turns visual and language input into robot actions. The ER model supports reasoning and planning; the on-device model emphasizes local execution.
Google DeepMind · Publication date not disclosed · Source accessed 2026-09-28
Official page read during live research; confirms identity and listed offering. Performance language remains a company claim.
Access and evaluation
The family page links model-specific access and research information. A demonstration on a partner robot does not establish unrestricted model availability or general production readiness.
Related reading is an editorial crosslink. Sourced connections describe relationships reported in the cited material. A link to a versioned profile does not establish compatibility with that version unless the connection note explicitly identifies it.