The industry's dominant search engine operator has launched an interface tailored exclusively for rapid evaluation workloads, taking direct aim at the cost overhead of standard generative calls. Google Deepmind is introducing the Gemini Decisions API, a dedicated interface designed to resolve closed-ended tasks such as yes/no determinations, multiple-choice questions, and scale-based ratings.
Rather than forcing applications to pay generation costs for full text completion, the underlying Gemini Decision 1 architecture evaluates candidates directly. The model processes up to 1,000 independent questions in a single pass across text, images, audio, video, and PDFs, returning probabilities alongside confidence scores within a window of 150 to 250 milliseconds.
Internal Workloads and Commercial Distribution
Google reports that it has already integrated this architecture across several internal operational pipelines. The company states it relies on the API for dynamic model routing inside Gemini Notebook, design prototyping in Stitch, and safety checks throughout Google Search.
The API costs $0.10 per million input tokens with no output-token charge.
At this price point, Google calculates that processing is approximately five times cheaper than utilizing Gemini 3.5 Flash-Lite. Access is available immediately via Google AI Studio, while third-party ecosystem integrations are rolling out across OpenRouter, LangChain, Vercel, and other platforms.
Competitive Realities Across Infrastructure
This pricing model lands amid a rapid repositioning across the machine learning sector away from generic text streaming toward purpose-built routing components. Market movements started when startup Jev introduced its version, triggering immediate counter-releases as OpenAI, Cloudflare, and Microsoft followed with similar offerings.
For enterprise architectures reliant on continuous categorization, the risk centers on vendor lock-in as foundational providers bundle classification layers directly into their hosted developer consoles. Teams that build around these hyper-specialized decision endpoints commit their routing topologies to proprietary provider scoring metrics, balancing immediate operational expenditure reductions against long-term portability.
Side by side
Data sorting and routing expenses drop sharply as Gemini Decisions API cuts out the cost of full text generation entirely. Immediate operational savings on data routing end up locking backend pipelines into one vendor's proprietary scoring metrics.