On November 11, 2026, Google launched the Gemini Decisions API alongside its specialized gemini-decision-1 model, available in Google AI Studio and the Gemini Developer API v1beta. Unlike conventional language models, this API generates no text: its sole function is to answer yes/no, multiple-choice, or rubric questions against a shared context, returning calibrated probabilities for each answer. (Source: MarkTechPost, November 11, 2026.)
What Is the Gemini Decisions API?
The API accepts a single multimodal context of up to 262,144 tokens (256K), which can include text, images, audio, video, and PDFs. Against that context, developers can submit up to 1,000 independent questions in a single call; the API scores each one and returns an estimated probability. The model bills input tokens once regardless of question count, making it considerably more cost-effective for large batches. Input pricing is $0.10 per million tokens; output tokens are not charged. Google states the per-task cost is approximately five times lower than using Gemini 3.5 Flash-Lite through the Interactions API on the same evaluation suite.
"An API that structures the decision instead of generating free-form text represents a paradigm shift for business automation: more predictable, more auditable, and in many cases less expensive."
Davarion Group & LabsWhat Changes for SMBs
- 01Automatic classification of tickets, support requests, or emails into categories without generating long text responses — a single call evaluates hundreds of records at once.
- 02Document review with multimodality: PDF contracts, product images, or call recordings can be processed in the same context to validate business criteria.
- 03Approval automation: workflows such as 'Does this invoice meet requirements?' or 'Is this form complete?' can be resolved with high accuracy at a cost per decision potentially below a fraction of a cent.
- 04Key limitation: the API has no open weights, runs only on Google's cloud, and does not support context caching or the Batch API in this beta, making it unsuitable for strictly confidential data that cannot leave the premises.
The $0.10 per million input token price places Decision 1 in the same range as OpenAI's decision model, according to industry sources, although TypeSafe Jev remains the cheapest option on the market (approximately $0.042/M tokens). For SMBs already using the Google ecosystem (Google Workspace, BigQuery), native integration and Google AI Studio's free tier significantly lower the adoption barrier.
At Davarion Group & Labs, we evaluate new AI APIs to identify which ones deliver real value for small and medium-sized business operations in Houston. If your business processes large volumes of documents, requests, or records and you want to explore whether the Gemini Decisions API can reduce manual work, reach out for a no-cost evaluation.
