~/AI/google-introduces-gemini-decisions-api-for-multimodal-decision-making

Google Introduces Gemini Decisions API for Multimodal Decision-Making

Google has launched the Gemini Decisions API with its 'Gemini Decision 1' model, entering the emerging decision model category alongside Jev and OpenAI. Unlike competing models that are limited to text, JSON, or images, Gemini Decision 1 offers full multimodality supporting text, images, audio, video, and PDFs. This release expands the growing landscape of specialized decision-making models by introducing native video and audio processing capabilities to structured probability outputs. Priced competitively at 10 cents per million input tokens with free output, it gives developers a powerful, cost-effective tool for automated evaluation and classification workflows. Gemini Decision 1 supports three core question types—yes/no, choice, and rubric-scoring—reusing terms like 'Noul' for yes/no queries. When handling video inputs, unlike Jev and OpenAI which output absolute 1.0 or 0.0 binary confidence scores, Gemini returns granular probabilistic floating-point values.

## BACKGROUND

Traditional Large Language Models (LLMs) generate free-form text responses which require parsing for programmatic use. Recently, a new category of decision models emerged to directly output structured answers with confidence scores and probabilities, streamlining automated decision workflows and evaluations.

## REFERENCES

## KEYWORDS

#AI#LLM#Google Gemini#APIs#Multimodal AI

$ subscribe --daily