Gemini Robotics
ER 1.6
Model Card
                            Gemini Robotics-ER 1.6 - Model Card

Model Cards are intended to provide developers with essential, summarized information on
models, including overviews of known limitations and mitigation approaches. Model cards may be
updated from time to time; for example, to include updated evaluations as the model is improved
or revised. The model card below covers Gemini Robotics-ER 1.6.

Published: April 2026

                                          Model Information

Description: Gemini Robotics-ER (Embodied Reasoning) 1.6 is a Vision-Language-Model that enhances
Gemini’s spatial and physical reasoning capabilities. It is based on Gemini 3.0 Flash.

Inputs: Text strings (e.g., a question, a prompt, document(s) to be summarized), images, audio, and
video files, with a token context window of up to 128k.

Outputs: Text, with a 64K token output.

Architecture: Architecture: Gemini Robotics-ER 1.6 is based on the Gemini 3.0 model family. For more
information about the model architecture, see the Gemini 3.0 Flash model card.

                                             Model Data

Training Dataset: Gemini Robotics-ER 1.6 was trained on Gemini 3.0 training datasets and additional
datasets representing various embodied reasoning tasks. For more information about the training data,
see the Gemini 3.0 Flash model card.​

Training Data Processing: For more information about the training data processing, see the Gemini 3.0
Flash model card.
                                Implementation and Sustainability

Hardware: Gemini Robotics-ER 1.6 was trained using Google’s Tensor Processing Units (TPUs). See
details in the Gemini 3.0 Flash model card.

The efficiencies gained through the use of TPUs are aligned with Google's commitment to operate
sustainably.

Software: Software JAX (Bradbury et al., 2018), ML Pathways (Dean, 2021).

                                                 Evaluation

Approach: See Figure1, Figure 2 in the release post for Gemini Robotics-ER 1.6 evaluation procedures
and Figure 3 for Gemini Robotics Safety evaluation procedures.

Results: See Figure 1, Figure 2 in the release post for Gemini Robotics-ER 1.6 evaluation results, and
Figure 3 for Gemini Robotics Safety evaluation results.

                                 Intended Usage and Limitations

Benefit and Intended Usage: Gemini Robotics-ER 1.6 represents an advancement to our reasoning-first
Gemini Robotics-ER models that enables robots to understand their environments with precision. This
model specializes in reasoning critical for robotics, including visual and spatial understanding, task
planning and success detection.

Known Limitations: For more information about the known limitations for Gemini Robotics-ER1.6, see
the Gemini 3.0 Flash model card.

Acceptable Usage: For information about the acceptable usage for Gemini Robotics-ER1.6 see the
Gemini 3.0 Flash model card. Additionally, users are required to use discretion before using the Robotics
Models in a production, commercial, or public environments, and to not use the Robotics Models for
safety-critical applications or work, such as in the following settings: (i) healthcare, (ii) transportation, or
(iii) other areas where safety protocols are vital, and a malfunction could reasonably foreseeably lead to
death, personal injury, or property damage.
                                        Ethics and Safety

Ethical Considerations & Risks: Previous impact assessment and risk analysis work as discussed in
(Gemini ER1.5 report) and references therein remain relevant to Gemini Robotics. See Section 6 for
information on responsible development and safety mitigations

Evaluation: Gemini Robotics-ER 1.6 demonstrates compliance with Gemini safety policies on adversarial
spatial reasoning tasks, and shows a substantially improved capacity to adhere to physical safety
constraints compared to previous generations. To ensure real-world readiness, we tested how well the
model identifies safety hazards in human-centric scenarios.