AI News

Google Launches Gemini Decisions API, Powered by the Gemini Decision 1 Model

English text is being prepared.Updated Nov 12 15:15

AI summary

Google announced the launch of the Gemini Decisions API, powered by the all-new Gemini Decision 1 model, which can complete up to 1,000 yes/no judgments, option selections, and criteria evaluations on text, image, video, and PDF content in a single fast pass. The API targets decision tasks such as tool selection, policy checks, model routing, and context quality evaluation, supports up to 256K Tokens per input, and has a latency of 150 to 250 milliseconds. Input pricing is $0.10 per 1 million Tokens, output Tokens are free of charge, and it is currently available in Google AI Studio with a free quota.

Why read this

AI score 69

The original text provides the latency and pricing for completing a thousand judgments in a single forward pass, so readers can use it to judge which types of decision tasks it suits.

Topics

Tags

Body · AI translation

Google Launches Gemini Decisions API for Fast Multimodal Decision-Making, Powered by the Gemini Decision 1 Model

English text is being prepared.

English text is being prepared.

English text is being prepared.

According to the introduction, the Gemini Decisions API is purpose-built for decision tasks such as tool selection, policy checks, model routing, and context quality assessment. Developers can input one piece of multimodal content containing text, images, audio, video, and PDFs (supporting up to 256K tokens) and pose up to 1,000 mutually independent questions about that content.

English text is being prepared.

In addition, the API is built on Gemini's native multimodal architecture. Developers can combine content of different modalities in the same input, for example pairing a call recording with account metadata, or placing a scanned PDF insurance claim document together with photos of the damage. They can also adjust the resolution of image and video inputs to control visual detail and token costs.

The API is currently available in Google AI Studio, where developers can get free quota for prototyping.

Source

English text is being prepared.