GOOGLE DEEPMIND (GOOGLE) · MODEL RELEASE TRACKER
gemini-decision-1
Gemini Decision 1 (gemini-decision-1) is a purpose-built decision model from Google that powers the Gemini Decisions API. It scores closed-set answers (yes/no, multiple choice, rubric) across multimodal inputs in a single forward pass, returning probabilities and confidence scores with low latency and a per-input-token pricing model.CURRENT SNAPSHOT5/5 DIMENSIONS WITH DATA
The dimensions that change the decision.
PricingPriced at $0.10 per 1M input tokens with zero output tokens billed.
PricingPriced at $0.10 per 1M input tokens with zero output tokens billed; Google states the Decisions API with Gemini Decision 1 is nearly 5× cheaper across the shared benchmark suite than Gemini 3.5 Flash-Lite (Interactions API).
Input / context sizeAllows a multimodal input up to 256K tokens; the documentation also reports scoring across multimodal contexts up to 16K tokens.
Context size for reported model computeReported model compute and latency figures are described for multimodal inputs and contexts up to 16K tokens.
Supported modalitiesProcesses text, images, audio, video, and PDFs directly in a single forward pass and can combine modalities in one shared input.
Benchmark performanceOn benchmarks spanning agent routing, reasoning, ranking, and multimodal understanding, Gemini Decision 1 led across the suite and exhibited the lowest calibration error.
AvailabilityAvailable through the Gemini API in Google AI Studio (reported as available now).
RELEASE TIMELINE
Published, source-backed release events only.
VERIFIABLE FACTS
Every value stays attached to a source and date.
RELEASE · AnnouncementDEVELOPER CLAIM
Announced by Google as the model powering the Gemini Decisions API (gemini-decision-1).
AVAILABILITY · AvailabilityDEVELOPER CLAIM
Available through the Gemini API in Google AI Studio (reported as available now).
CAPABILITIES · Question-scoring capabilityDEVELOPER CLAIM
Can evaluate up to 1,000 independent yes/no, multiple-choice, and rubric-scoring questions in a single forward pass over a shared multimodal input.
CAPABILITIES · Scoring method and latencyDEVELOPER CLAIM
Scores candidate options in a single forward pass without token-by-token decoding; returns probabilities and confidence scores with reported end-to-end latency of about 150–250 ms and median model compute of 50–100 ms.
MODALITIES · Supported modalitiesDEVELOPER CLAIM
Processes text, images, audio, video, and PDFs directly in a single forward pass and can combine modalities in one shared input.
CONTEXT WINDOW · Input / context sizeDEVELOPER CLAIM
Allows a multimodal input up to 256K tokens; the documentation also reports scoring across multimodal contexts up to 16K tokens.
CONTEXT WINDOW · Context size for reported model computeDEVELOPER CLAIM
Reported model compute and latency figures are described for multimodal inputs and contexts up to 16K tokens.
PRICING · PricingDEVELOPER CLAIM
Priced at $0.10 per 1M input tokens with zero output tokens billed.
CAPABILITIES · Latency / performanceDEVELOPER CLAIM
Returns probabilities and confidence scores in about 150–250 ms end-to-end, with median model compute reported at ~50–100 ms.
BENCHMARKS · Benchmark performanceDEVELOPER CLAIM
On benchmarks spanning agent routing, reasoning, ranking, and multimodal understanding, Gemini Decision 1 led across the suite and exhibited the lowest calibration error.
PRICING · PricingDEVELOPER CLAIM
Priced at $0.10 per 1M input tokens with zero output tokens billed; Google states the Decisions API with Gemini Decision 1 is nearly 5× cheaper across the shared benchmark suite than Gemini 3.5 Flash-Lite (Interactions API).
WHAT CHANGED
Stored passport versions, without reconstructed history.
RELEASE · AnnouncementIntroduced in a Google blog post as the new Gemini Decision 1 model powering the Gemini Decisions API.→Announced by Google as the model powering the Gemini Decisions API (gemini-decision-1).
AVAILABILITY · AvailabilityAvailable through the Gemini API in Google AI Studio.→Available through the Gemini API in Google AI Studio (reported as available now).
BENCHMARKS · Benchmark performance+ On benchmarks spanning agent routing, reasoning, ranking, and multimodal understanding, Gemini Decision 1 led across the suite and exhibited the lowest calibration error.
PRICING · Pricing+ Priced at $0.10 per 1M input tokens with zero output tokens billed; Google states the Decisions API with Gemini Decision 1 is nearly 5× cheaper across the shared benchmark suite than Gemini 3.5 Flash-Lite (Interactions API).
CONTEXT WINDOW · Input / context size+ Allows a multimodal input up to 256K tokens; the documentation also reports scoring across multimodal contexts up to 16K tokens.
CAPABILITIES · Latency / performance+ Returns probabilities and confidence scores in about 150–250 ms end-to-end, with median model compute reported at ~50–100 ms.
CONTEXT WINDOW · Maximum multimodal input tokens (stated)− The Decisions API accepts a multimodal input of up to 256K tokens when passing inputs and questions to the model.
MODALITIES · Supported modalities+ Processes text, images, audio, video, and PDFs directly in a single forward pass and can combine modalities in one shared input.
MODALITIES · Supported modalities− Built on Gemini’s native multimodal architecture and can process text, images, audio, video, and PDFs directly in a single forward pass; modalities can be combined in one shared input.
CAPABILITIES · Question-scoring capabilityCan evaluate up to 1,000 independent yes/no, multiple-choice, and rubric-scoring questions in a single request and returns probabilities and per-answer confidence scores (0–1).→Can evaluate up to 1,000 independent yes/no, multiple-choice, and rubric-scoring questions in a single forward pass over a shared multimodal input.