Introducing the Gemini Decisions API for structured model outputs
Google DeepMind has introduced the Gemini Decisions API, powered by the new Gemini Decision 1 model.
2 min read, Thursday, 12 November

Google DeepMind has introduced the Gemini Decisions API, a new endpoint designed for fast, closed-set decisions. Powered by a new model called Gemini Decision 1, the system evaluates up to 1,000 independent questions in a single forward pass across multiple types of input data, including text, images, audio, video, and PDF documents.
Key facts
- Model
- Gemini Decision 1
- Max questions per request
- 1,000
- Max context window
- 256K tokens
- End-to-end latency
- 150-250 ms
- Input price
- $0.10 per 1M tokens
- Output price
- $0 per 1M tokens
Fast multimodal evaluation
Unlike traditional text generation models that write answers token by token, Gemini Decision 1 scores candidate options directly. This approach yields an end-to-end latency of 150 to 250 milliseconds. The architecture supports contexts up to 256K tokens, though median latencies of 50 to 100 milliseconds apply for contexts up to 16K tokens.

The API supports three decision types:
- noul (Yes/No): Estimates the probability of a binary condition with optional true and false description criteria.
- choice (Multiple Choice): Selects from 2 to 255 choices, returning the winning option and normalized probabilities.
- score (Ordered Rubric): Grades inputs across an ordered rubric starting at 0, returning a continuous weighted score.
Every answer includes a confidence score ranging from 0 to 1, enabling applications to route ambiguous cases to fallback logic or human reviewers.
Pricing and availability
Gemini Decision 1 is priced at $0.10 per 1 million input tokens, with zero output tokens billed. Developers can access the API in Google AI Studio using Python, JavaScript, and REST guides. Integrations are rolling out across platforms including LangChain, Vercel, Pydantic AI, and OpenRouter.
Source: Original announcement
Worth a read?



Comments