Gemini Decisions API

Google has released the Gemini Decisions 1 API, a decision‑making model that joins the existing Jev and OpenAI offerings. Like Jev, Gemini supports three query types: yes/no, option selection, and rubric‑based scoring, and it even reuses the term “Noul” for yes/no questions. Its pricing matches OpenAI’s GPT‑6 Luna model, at $0.10 per million input and output tokens with no extra fee; Jev remains the cheapest at $0.042 per million.

Unlike its competitors, Gemini Decisions 1 is fully multimodal: it can process text, images, audio, video, and PDF documents. In a practical test, the author uploaded a video of a robot walking through Yonkers and asked about the presence of a dragon, a robot, and the geographic location of the content. While Jev and OpenAI returned binary answers (0 or 1), Gemini provided intermediate probabilities, showing greater nuance in its confidence.

This development matters because it expands the capabilities of automated decision systems, enabling developers to integrate more sophisticated multimedia content analysis without output costs. That could accelerate AI adoption in applications that require quick, accurate evaluations.

Read the original article on Simon Willison

This summary is an informational synthesis produced by dataqbs.com. All rights to the original content belong to its author and the cited media outlet. We act solely as curators of technology news and claim no authorship.