Gemini Decision 1 has the highest scores among decision models on the knowledge and reasoning evaluations in the Decision Index. On AA-Omniscience-Decision it scores 66, ahead of d1-ultra (64) and GLiDE v1.2 (61) and level with GPT-6 Luna (Max) (66). It answers 45% of questions correctly, compared with 28% for Drex Max v1.1 and 27% for GLiDE v1.2
On HLE-MultiChoice it scores 49, ahead of Mammoth 27B and d1-ultra (both 47). Its accuracy is in line with Decider v1.2 and Jev 1.14, but it leads on Decision Score due to calibration: its Brier score of 0.20 is among the lowest of all models on the evaluation
