Google offers a decisions API that directly returns calibrated probabilities instead of text. Announced at $0.10 per million input tokens and given for 150 to 250 ms of end-to-end latency, it arrives without caching or batching and remains more expensive per token than Jev.
No cache or batch, and a higher entry cost than Jev
The API does not yet offer prompt caching or batch API, which means that any repeated context is rebilled in its entirety. At $0.10 per million input tokens, Gemini Decision 1 remains more expensive than Jev 1.14 posted at $0.042 per million. In the comparison table, caching is shown as unsupported on Google side, supported on OpenAI Decisions, and undisclosed on Jev. On the latency side, Google announces 150 to 250 ms end-to-end, while OpenAI Decisions is described as around ten times faster than its Responses API, with no equivalent figure published for Jev.
Internal uses and ecosystem in deployment
Google reports internal use in Gemini Notebook to route to different model levels, with up to 40% less latency on daily queries. Google Labs uses Creative Chips to select emoji reactions from more than 200 possibilities. The API is used by Stitch’s Quick Canvas to generate playable mockups in seconds. Support across the ecosystem is currently being built into OpenRouter, LangChain, Vercel AI SDK, LiteLLM, Pydantic AI, Pipecat, LiveKit, and OpenCode, with support in Gemini on Google Cloud said to be coming.
A multimodal decision API that returns calibrated scores
Google launches Gemini Decisions, based on the specialized Gemini Decision 1 model, which does not generate text and directly returns calibrated probabilities. The model is multimodal and processes text, image, audio, video and PDF in a single pass, with answers accompanied by a confidence between 0 and 1. Each query groups together a model ID, a shared input and a question card, these being evaluated in parallel and without mutual interference. Three question formats are provided: yes/no with probability bounded between 0.01 and 0.99, single choice from 2 to 255 options with probabilities per option, and score from 2 to 255 ordered levels. The API processes up to 1000 questions per call with 0 output tokens. Google recommends redirecting low trust cases to a backup plan or human review.
Deciphered AI dev tools, agents and new features, every evening in 5 minutes. Free.
Included upon registration: our selection of the best AI guides & comparisons.
Choose your pace
Free · No spam · Unsubscribe in 1 click
Announced latency and billing focused on shared input
Google reports a median calculation time of 50-100 ms and end-to-end latency of 150-250 ms, which applies to inputs up to 16K tokens and remains nearly stable for a number of questions ranging from 1 to 100. Shared input is only charged once per query, with each question charging only for its own instruction tokens. and criteria, at no additional cost for reflection, transcription or OCR tokens. The available context capacity is 262,144 tokens for shared input and questions. The service is accessible only via the Gemini Developer API in v1beta, charged $0.10 per million input tokens, and supports up to 1,000 questions per request.
Self-declared benchmarks and measured calibration
According to figures provided by Google, Gemini Decision 1 obtains the best performance on 7 of the 9 published accuracy benchmarks, with the lowest calibration error. The model achieves 92.2% on JevBench and records its lowest Macro‑F1 at 58.7% on Amazon ESCI. It is 79.9% on When2Call compared to 82.0% for Jev 1.14. Other reported scores include 79.7% on New Yorker Caption, 86.2% on Mind2Web Multimodal, 79.8% on PubMedQA, 86.0% on Global-MMLU-Lite, 71.8% on MMAU, and 69.1% on Video-MME without captions. The Top‑1 ECE error is given as 0.0250, about half that of Jev 1.14 at 0.0494. For seven shared reviews, the total price quoted is $0.88 for Gemini Decision 1 versus $0.68 for Jev 1.14, $0.93 for GPT-6 Luna, and $4.14 for Gemini 3.5 Flash-Lite. Google also claims that Gemini Decision 1 is almost five times cheaper than Flash-Lite on this suite.
Developer access and extensive comparisons
The endpoint is active on Gemini Developer API v1beta and requires google‑genai SDK 3.2 or higher, with a cost-free tier to prototype in Google AI Studio and guides for Python, JavaScript, and REST. In published comparisons, OpenAI Decisions shows a context of 1,050,000 tokens on its model page, while Jev 1.14 shows 64K per request, expandable to 262K with opt-in header. Gemini Decision 1 accepts 1 to 1,000 questions per query compared to 1 to 400 for OpenAI Decisions, and Jev is context limited with no numerical cap. The input modalities listed range up to PDF, video and audio for Gemini Decision 1, where OpenAI Decisions quotes text, image and JSON, and Jev text. The entry prices referenced are $0.10 per million for Google, $0.10 uncached for OpenAI, and $0.042 for Jev. The Google API remains reserved for Google hosting, without self-hosting.

