Gemini Decision 1 now available on AI Gateway
Also covered by 3 sources
What happened
Google released the Gemini Decision 1 model through its Gemini Decisions API on November 11, 2026. The model is accessible via AI Gateway and allows users to submit up to 1,000 questions per request using various input types, including text, images, audio, video, and PDFs. The API returns probabilistic and choice-based outputs for decision-making tasks without generating output tokens. The service is available through a Google-hosted API and is priced based on input tokens.
Key details
- Supports up to 1,000 independent questions per request.
- Handles text, image, audio, video, and PDF inputs in a single pass.
- Returns confidence scores for yes/no, fixed-choice, and rubric questions.
- Priced at $0.10 per million input tokens, with no charge for output tokens.
- Context window is 262,144 tokens per request.
Why it matters
Developers and companies can now automate complex decision-making tasks across multiple data types with lower latency and cost per request. This enables more efficient routing, policy checks, and grading within software agents, with no output token charges. Teams using the API must consider the lack of prompt caching and batch API, which affects repeated or large-scale requests.