The Model Gap

Google DeepMind

Gemini Decision 1

Gemini Decision 1 launched on 2026-11-11 as gemini-decision-1, a Decisions API model that returns calibrated probabilities and does not generate text. Paid Standard is $0.10 per million input tokens. Output is not applicable, because the run emits 0 output tokens. The model page states a 262,144-token input limit. A Free tier exists. No tracked score is stored, Argon and Carbon stay current, and this is not GPT-6 Luna.

Gemini Decision 1 benchmarks and pricing, every number sourced: no Gemini Decision 1 benchmark scores tracked yet, priced at $0.10 per million input tokens and $0.00 per million output.

Released
2026-11-11
License
proprietary
Context window
262K tokensvendor-stated (aistudio.google.com/docs/models/gemini-decision-1: input token limit 262,144; checked 2026-11-12)
Knowledge cutoff
Not disclosed
Parameters
Not disclosed
Architecture
Not disclosed

Gemini Decision 1 API pricing

$0.10 in / $0.00 out per 1M tokens — official pricing source

What Gemini Decision 1 costs per job

Gemini Decision 1 cost for three reference workloads, computed from its list rates
WorkloadTokens in / outCost
One long chat turn100K / 10K$0.010
A codebase review1,000K / 100K$0.100
A day of agent work10,000K / 1,000K$1.00

Computed from Gemini Decision 1’s list rates above — cache discounts and batch tiers are not applied.

Gemini Decision 1 is one of 7 Google DeepMind models tracked on this site, at these official list prices.

Google DeepMind model pricing, official list rates
ModelIn / 1MOut / 1M
Gemini 4 Hydrogen$1.50$7.50
Gemini Decision 1$0.10$0.00
Gemini 4 Carbon$4.00$20.00
Gemini 4 Argon$4.00$20.00
Gemini 3.8 Flash$0.75$3.75
Gemini 3.7 Flash(previous version)$0.75$3.75
Gemini 3.1 Pro Preview$2.00$12.00

Notes on the record

Google's blog dated November 11, 2026, written by Min Choi and Felix Yu of Google DeepMind, introduces gemini-decision-1 for the Gemini Decisions API. The quickstart is API version v1beta in Google AI Studio. The post says Google Cloud support is coming soon. This page does not call Vertex or Cloud generally available.

The model page checked November 12, 2026 calls it a specialized Decisions API model that does not generate text. It accepts text, image, video, audio, and PDF. Output is calibrated probabilities, and the output token limit is 0. One request can carry up to 1,000 questions. Text generation, thinking, caching, and the Batch API are not supported.

Paid Standard is $0.10 per million input tokens. The pricing page marks output as not applicable because the model emits 0 output tokens, so the stored output rate is 0. The Free tier lists input as free of charge and that figure is not the paid headline. Gemini 4 Carbon's pricing row, checked November 11, 2026, had no Free tier. Free-tier content may be used to improve Google's products. Paid content may not. Shared input is billed once.

The model page states an input limit of 262,144 tokens. The blog's up to 256K is rounding, so this page stores 262,144 and not 256,000. Knowledge cutoff, parameter count, architecture, and license are unpublished. The license is recorded as proprietary. The family is Gemini Decision, not Gemini 4, so Argon and Carbon are not marked superseded.

These lines are vendor claims, not scores. Google says Decision 1 leads a suite spanning agent routing, reasoning, ranking, and multimodal understanding, with the lowest calibration error, and is nearly 5x cheaper on that shared suite than Gemini 3.5 Flash-Lite with the Interactions API. It states 50 to 100 ms of model compute and 150 to 250 ms end to end, for multimodal inputs and contexts up to 16K tokens, nearly flat from 1 to 100 questions.

Compare with

Gemini 4 HydrogenGemini 4 CarbonGemini 4 ArgonGemini 3.8 Flashany model…

FAQ

What does Gemini Decision 1 cost?

Paid Standard is $0.10 per million input tokens. The pricing page marks output as not applicable, because the model emits 0 output tokens, and this page stores that rate as 0. The Free tier lists input as free of charge, and that figure is not the paid headline. Free-tier content may be used to improve Google's products. Paid content may not. Caching and the Batch API are unsupported, so there is no second paid rate. Shared input is billed once, and each question adds its own tokens. The per-job table multiplies the stored $0.00 by a fixed output assumption. That is not a claim this model writes those tokens. Checked November 12, 2026.

Does Gemini Decision 1 generate text?

No. It is a Decisions API model. One request can ask up to 1,000 yes/no, choice, or rubric questions and returns calibrated probabilities, with an output token limit of 0. It accepts text, image, video, audio, and PDF. The documented API version is v1beta in Google AI Studio. The launch post says Google Cloud support is coming soon, which this page does not read as general availability.

What is Gemini Decision 1's context window?

The model page checked November 12, 2026 states an input token limit of 262,144. The November 11 blog rounds the same limit as up to 256K. This page stores 262,144 and does not store 256,000. The output token limit is 0. No knowledge cutoff is published.

Does Gemini Decision 1 replace Gemini 4 Argon or Carbon?

No. Its family is Gemini Decision, not Gemini 4, so the previous-version rule does not attach it to Argon or Carbon, and neither page is marked superseded. It is also not a GPT-6 Luna equivalent. Luna generates text at $0.10 input and $0.50 output. Decision 1 does not generate text, and its paid output cell is 0 tokens.

Has Gemini Decision 1 been independently benchmarked?

No tracked board had a Gemini Decision 1 row at the November 12, 2026 scan, so this page stores no score. The launch post's suite, calibration, and cost lines stay vendor claims: a lead across agent routing, reasoning, ranking, and multimodal understanding, the lowest calibration error, and nearly 5x cheaper than Gemini 3.5 Flash-Lite with the Interactions API on that shared suite. The latency claim is 150 to 250 ms end to end, and only for multimodal inputs and contexts up to 16K tokens.

What are Gemini Decision 1's free rate limits?

On the rate-limits page checked November 12, 2026, Free is 200 requests per minute, 50,000 tokens per minute, and 10,000 requests per day. Paid Tier 1 is 1,500 requests, 1,000,000 tokens, and 50,000 requests. Tier 2 is 5,000, 4,000,000, and unlimited. Tier 3 is 15,000, 10,000,000, and unlimited. The model page marks the Batch API and priority inference as not supported, so those general quota sections are not a Decision 1 offering.

Further reading