Gemini Robotics-ER 1.6
2026-04-14 · Publisher : Google · Domain : Artificial Intelligence
Google DeepMind publie Gemini Robotics-ER 1.6, modèle de raisonnement incarné amélioré en compréhension spatiale et multi-vues, capable de lire jauges et voyants, disponible via l'API Gemini.
What changes
- Meilleure compréhension spatiale et multi-vues, planification de tâches et détection de succès.
- Nouvelle capacité de lecture d'instruments (jauges complexes, voyants de niveau), développée avec Boston Dynamics.
- Déployé dans le robot Spot de Boston Dynamics pour des inspections industrielles.
- Présenté comme le modèle de robotique le plus sûr de Google à cette date.
Google DeepMind releases Gemini Robotics-ER 1.6, an embodied reasoning model improved in spatial and multi-view understanding, able to read gauges and indicator lights, available through the Gemini API. ### What changes - Better spatial and multi-view understanding, task planning and success detection. - New instrument-reading capability (complex gauges, level indicators), developed with Boston Dynamics. - Deployed in Boston Dynamics' Spot robot for industrial inspections. - Presented as Google's safest robotics model at that date. ### Specifications - Availability: Gemini API and Google AI Studio from April 14, 2026. - API identifier (changelog): gemini-robotics-er-1.6-preview, shutdown announced for August 31, 2026 in favor of version 2. - Context, outputs and knowledge cutoff: not disclosed in the blog post. ### Pricing Pricing not disclosed at launch. ### Benchmarks No figure published in the blog post consulted. ### Key takeaways A reasoning model for robots already in production at a partner, replaced three months later by Gemini Robotics ER 2.
Links
Tags : LLM, Google, Gemini, Robotique, Raisonnement incarné