Google Launches Gemini Decision 1 at 2.4x Microsoft’s Price

Google launched Gemini Decision 1 on 11 November, a model that scores closed-set questions instead of writing text, priced at $0.10 per million input tokens with “zero output tokens billed”, according to Google’s blog post. Microsoft-Decision-1 lists $0.042, so Google’s price is about 2.4 times Microsoft’s, by our arithmetic.

The model is served through a new Gemini Decisions API in Google AI Studio. Sundar Pichai announced it on X at 18:15 UTC, about 89 minutes after the first tracker post we found listing the model name, by our arithmetic. We reported that sighting in our earlier piece, when Google had not announced it.

Business Pill · Usage-Based Pricing

An evergreen Business Engineer explainer on usage-based pricing: paying for work, not for seats. It teaches the general idea only and does not discuss Gemini, Microsoft or decision models.

What Google Launched

Google’s post says the Decisions API gives developers “purpose-built primitives for bounded choices like tool selection, policy checks, model routing, and context grading”. One request takes a multimodal input of up to 256K tokens and up to 1,000 independent questions.

Rather than generating answer text, the post says, “Gemini Decision 1 scores your candidate options in a single forward pass, returning probabilities and confidence scores in 150-250 ms end-to-end latency at $0.10 per 1M input tokens with zero output tokens billed”.

The model page lists three question types. A yes/no question “Returns a calibrated probability from 0.01 to 0.99 and a confidence score”; single-select and rubric questions take 2 to 255 options or levels. It lists an input token limit of 262,144 and an output token limit of 0.

What Was Posted on X

Logan Kilpatrick of Google introduced Gemini Decision 1 in the Gemini API at 18:16 UTC, giving the $0.10 per 1M input token price, native image, audio, video and PDF support, a 256K context window and 1K questions per request.

Google’s own account opened a thread on the Gemini Decisions API at 18:26 UTC, saying the model evaluates one input against up to 1,000 questions in a single pass with zero output tokens billed.

Sundar Pichai’s launch post at 18:15 UTC, the first official post we found, lists three decision types, the $0.10 price and his ‘~5x cheaper than Flash-Lite’ claim, which Google’s blog frames as cost across its benchmark suite.

The Price on Google’s Own List

Google’s pricing page lists Gemini Decision at $0.10 per million input tokens for text, image, video, audio and PDF, with output “Not applicable (0 output tokens)”. It says each request “bills shared input tokens once, regardless of question count”.

On the same page, Gemini 3.5 Flash-Lite input costs $0.30, Gemini 3.8 Flash $0.75 and Gemini 4 Carbon $2.00 through 30 November. By our arithmetic, those are 3, 7.5 and 20 times the decision model’s input price, before any output charges. Carbon’s price doubles on 1 December, as we covered in our Carbon launch piece.

Pichai wrote that the price is “~5x cheaper than Flash-Lite”. Google’s blog frames that comparison as cost “across the shared benchmark suite” against Gemini 3.5 Flash-Lite with the Interactions API, not as a list-price ratio.

How It Compares With Microsoft’s Decision Model

Microsoft’s launch post for Microsoft-Decision-1, introduced on 9 October, reads: “Input tokens cost $0.042 USD per million tokens. Output tokens are free.” Microsoft says the model is “designed for routing, classification, prioritization, verification, and workflow control”.

Both companies bill only input. On list price, Microsoft’s model is cheaper per token; Google’s launched 33 days later, by our arithmetic. Microsoft says it “post trained Qwen3.5-9B” to build its model; Google’s post says its model is “Built on Gemini’s native multimodal architecture”.

What Google’s Benchmark Table Shows

Google’s blog says that on its benchmark suite, “Gemini Decision 1 leads across the suite, with the lowest calibration error”. The table is Google’s own, self-reported; we did not run the evaluations.

The table compares Gemini Decision 1 with TypeSafe Jev 1.14, OpenAI’s GPT-6 Luna on the OpenAI Decisions API, and Gemini 3.5 Flash-Lite. Against Jev, Google’s model scores higher on 5 of the 7 benchmarks where both have results, by our count, including 92.2% against 90.0% on JevBench.

Jev scores higher on the other two: 86.8% against 86.2% on Mind2Web Multimodal and 82.0% against 79.9% on When2Call. Google’s cost row puts the total price of the seven shared evaluations at $0.68 for Jev and $0.88 for Gemini Decision 1, about 29% more for Google’s model, by our arithmetic.

Who Is Using It

Google says teams inside the company already use the API. In Gemini Notebook, it says, the API routes queries to a model tier and serves “everyday queries with up to 40% lower latency”; Stitch, CC from Google Labs and Google Search also use it, according to the post.

The post says integrations are rolling out across OpenRouter, LangChain, Vercel, LiteLLM and others, with support “coming soon” to Gemini in Google Cloud. The Decisions API guide carries the request schemas.

The Structural Read

As we read Google’s pages, the product is the billing unit as much as the model: shared input is billed once per request, each question adds only its own tokens, and output costs nothing.

As we read the two launch posts, Google and Microsoft now sell the same kind of narrow scorer, and on list price Microsoft’s input token is cheaper, $0.042 against $0.10.

Three Implications

FOR DEVELOPERS Google says one request can carry up to 1,000 independent questions over a shared input of up to 256K tokens, across text, images, audio, video and PDFs.

FOR PRICING Google’s page lists Gemini Decision at $0.10 per million input tokens, against $0.30 for Gemini 3.5 Flash-Lite and $2.00 for Gemini 4 Carbon through 30 November.

FOR RIVALS Google’s own table puts the cost of the seven shared evaluations at $0.68 for TypeSafe Jev 1.14, $0.88 for Gemini Decision 1 and $0.93 for OpenAI’s GPT-6 Luna.

The Business Engineer Lens

This story maps onto the Business Engineer framework The State of the Inference Economy.

The framework says: “Inference is a revenue engine: continuous, distributed across millions of applications and agents, incurring marginal cost with every query.”

As we read it, a decision model prices that marginal cost for the narrowest query: Google lists $0.10 per million input tokens and Microsoft $0.042, both with output free. The framework does not discuss decision models or Gemini Decision 1.

What Is Not Established

Caveat: the benchmark scores, latency figures and the Notebook latency gain are Google’s own and have not been independently tested. The pricing page says the Batch API is not supported. The uses named in the post are Google’s own products. We did not contact Google or Microsoft for this piece.

The Bottom Line

Google launched Gemini Decision 1 at $0.10 per million input tokens with no output charges, about 2.4 times Microsoft-Decision-1’s $0.042, by our arithmetic. Google’s own table shows its model ahead of TypeSafe’s Jev 1.14 on 5 of 7 shared benchmarks, by our count, while listing Jev as cheaper to run them.