Google Launches Gemini Decision 1, a Production Decision Model
Google introduced Gemini Decision 1, a calibrated multimodal decision model for closed-set classification tasks, supporting up to 1,000 questions per request with 256K context and ~200ms end-to-end latency. Priced at $0.10 per million input tokens with no output cost, it claims to be 5x cheaper than Flash-Lite. Independent benchmarks show mixed results, with a Decision Index score of 66, tying Elephant 9B and Jev 1.14, but trailing Drex Max and GLiDE. Early adopters include Stitch and LangChain, with day-zero support.