GOOGLE DEEPMIND · MODEL RELEASE TRACKER
Gemini Decision 1
A purpose-built multimodal decision model (gemini-decision-1) for fast, closed-set scoring of yes/no, multiple-choice, and rubric questions across text, images, audio, video, and PDFs via the Gemini Decisions API.CURRENT SNAPSHOT5/5 DIMENSIONS WITH DATA
The dimensions that change the decision.
Pricing$0.10 per 1M input tokens; output tokens are not billed (zero output-token charge).
Maximum input tokensSupports multimodal inputs up to 256K tokens.
Context tokens (reported compute context)Reports operation across contexts up to 16K tokens (latency metrics reported at this context size).
ModalitiesProcesses text, images, audio, video, and PDFs directly in a single forward pass and can combine modalities in one shared input.
Supported modalitiesProcesses text, images, audio, video, and PDFs directly in a single forward pass and can combine modalities in one shared input.
Benchmark performanceLeads on benchmarks spanning agent routing, reasoning, ranking, and multimodal understanding, reporting the lowest calibration error across the suite.
AvailabilityAvailable through the Gemini Decisions API via the Gemini API in Google AI Studio; integrations (OpenRouter, LangChain, Vercel) are rolling out.
RELEASE TIMELINE
Published, source-backed release events only.
VERIFIABLE FACTS
Every value stays attached to a source and date.
RELEASE · Announcement and availabilityDEVELOPER CLAIM
Introduced with the Gemini Decisions API and available through the Gemini API in Google AI Studio; model identifier gemini-decision-1.
CAPABILITIES · Supported question types and max per requestDEVELOPER CLAIM
Can evaluate up to 1,000 independent yes/no, multiple-choice (choice), and rubric-scoring questions in a single request.
MODALITIES · 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 · Maximum input tokensDEVELOPER CLAIM
Supports multimodal inputs up to 256K tokens.
CONTEXT WINDOW · Context tokens (reported compute context)DEVELOPER CLAIM
Reports operation across contexts up to 16K tokens (latency metrics reported at this context size).
BENCHMARKS · Benchmark performanceDEVELOPER CLAIM
Leads on benchmarks spanning agent routing, reasoning, ranking, and multimodal understanding, reporting the lowest calibration error across the suite.
CAPABILITIES · Scoring method and latencyDEVELOPER CLAIM
Scores candidate options in a single forward pass (no token-by-token decoding), returning probabilities and confidence scores with 150–250 ms end-to-end latency and ~50–100 ms median model compute.
PRICING · PricingDEVELOPER CLAIM
$0.10 per 1M input tokens; output tokens are not billed (zero output-token charge).
AVAILABILITY · AvailabilityDEVELOPER CLAIM
Available through the Gemini Decisions API via the Gemini API in Google AI Studio; integrations (OpenRouter, LangChain, Vercel) are rolling out.
CAPABILITIES · Decision scoring capabilityDEVELOPER CLAIM
Scores up to 1,000 independent questions (yes/no, multiple choice, and rubric scoring) in a single forward pass, returning probabilities and confidence scores for each answer.
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.
CAPABILITIES · LatencyDEVELOPER CLAIM
Returns scores with ~50–100 ms median model compute time and ~150–250 ms end-to-end latency.
WHAT CHANGED
Stored passport versions, without reconstructed history.
AVAILABILITY · Availability+ Available through the Gemini Decisions API via the Gemini API in Google AI Studio; integrations (OpenRouter, LangChain, Vercel) are rolling out.
BENCHMARKS · Benchmark performanceOn benchmarks spanning agent routing, reasoning, ranking, and multimodal understanding, Gemini Decision 1 leads across the suite and has the lowest calibration error.→Leads on benchmarks spanning agent routing, reasoning, ranking, and multimodal understanding, reporting the lowest calibration error across the suite.
CONTEXT WINDOW · Context tokens (reported compute context)+ Reports operation across contexts up to 16K tokens (latency metrics reported at this context size).
PRICING · Pricing and billing− Priced at $0.10 per 1M input tokens with zero output tokens billed; stated to be nearly 5× cheaper on the shared benchmark suite than Gemini 3.5 Flash‑Lite with the Interactions API.
CAPABILITIES · Latency+ Returns scores with ~50–100 ms median model compute time and ~150–250 ms end-to-end latency.
CONTEXT WINDOW · Context for stated latency− Median compute latency figures (50–100 ms) and end-to-end latency (150–250 ms) are reported for inputs and contexts up to 16K tokens.
CONTEXT WINDOW · Maximum input tokens+ Supports multimodal inputs up to 256K tokens.
CONTEXT WINDOW · Max multimodal input tokens− Supports a multimodal input up to 256K tokens.
MODALITIES · Supported modalities+ Processes text, images, audio, video, and PDFs directly in a single forward pass and can combine modalities in one shared input.
PRICING · Pricing+ $0.10 per 1M input tokens; output tokens are not billed (zero output-token charge).
CAPABILITIES · Decision scoring capability+ Scores up to 1,000 independent questions (yes/no, multiple choice, and rubric scoring) in a single forward pass, returning probabilities and confidence scores for each answer.