Overview
Gemini Robotics-ER 1.5 is a preview vision-language model (VLM) optimized for embodied reasoning. It is designed to help robotics systems understand physical environments, plan multi-step tasks, and coordinate tools or robot actions. Google announced it on September 25, 2025. (deepmind.google)
Capabilities and uses
- Interprets visual scenes and video, reasons about spatial and temporal relationships, and can identify objects, their states and locations, and task progress or success. (developers.googleblog.com)
- Plans tasks from natural-language instructions and can use Google Search or developer-defined functions. It can act as a high-level orchestrator for a separate robot control system or the Gemini Robotics 1.5 vision-language-action model. Example applications include sorting objects and monitoring robot task progress. (deepmind.google)
- The model documentation lists text, image, video and audio inputs and text output, with limits of 1,048,576 input tokens and 65,536 output tokens. It lists function calling, code execution, structured output, search grounding and thinking as supported. (ai.google.dev)
Limitations and availability
- This is an embodied-reasoning VLM, not the separate Gemini Robotics 1.5 action model; physical execution requires integration with suitable robot control hardware and software. Its outputs and proposed actions should be validated, with independent safety measures appropriate to the robot and environment. (deepmind.google)
- The documented preview does not support audio or image generation, the Live API, context caching or Batch API. (ai.google.dev)
- Google’s deprecation documentation lists this endpoint among deprecated models, with April 30, 2026 as its listed shutdown date. The current recommended replacement is
gemini-robotics-er-2-preview. Availability should be checked before integration. (aistudio.google.com)
Pricing
As checked on November 18, 2026, an official price for this exact preview endpoint’s input and output usage could not be verified. Currency, rates and billing units are therefore unverified; this does not imply free use. The official Gemini API pricing page is the pricing reference. (ai.google.dev)