Google AI Launches Gemini Robotics-ER 1.5 preview, Cautions Users on Limitations & Privacy
MOUNTAIN VIEW, CA – Google AI for Developers has released Gemini Robotics-ER 1.5 into Preview, offering developers access to advanced robotics models via the Gemini API. However, the company simultaneously issued cautions regarding the model’s current status, potential limitations, and critical privacy considerations.
Gemini Robotics-ER 1.5, currently in Preview, is subject to API and capability changes and is not recommended for production-critical applications without extensive testing. Users may experience increased processing times with complex queries, high-resolution inputs, or extensive use of the thinking_budget parameter. Like other large language models, gemini Robotics-ER 1.5 can “hallucinate” and provide inaccurate facts, particularly with ambiguous or unfamiliar inputs. Output quality is directly tied to prompt clarity and specificity. Computational costs, especially with video inputs or high thinking_budget settings, are also a factor; details on “Thinking” are available on the Google AI for Developers site.
A key focus of the release is user duty regarding data privacy. Google emphasizes that the robotics Models utilize video and audio data to function and operate hardware. users are responsible for ensuring individuals are notified and consent to the potential collection of their Personal Data – including voice, imagery, and likeness – before interacting with or being present near the Robotics Models. The Gemini API Additional Terms of Service, found at https://ai.google.dev/gemini-api/terms, outline how Google uses this data. Google recommends employing techniques like face blurring and operating the models in areas without identifiable persons to minimize personal Data collection.
Detailed pricing information and regional availability can be found on the Google AI for Developers pricing page: https://ai.google.dev/gemini-api/docs/pricing.
