Models
Liquid AI Releases d1.2 and d1.2-ultra: Omnimodal Decision Models Topping the AA Decision Index
Liquid AI has launched d1.2 and d1.2-ultra, hosted decision models that return calibrated probabilities over fixed answer types instead of generated text, now accepting text, image, and audio input. d1.2-ultra tops the Artificial Analysis Decision Index v2.1 at 75 (d1.2 scores 69), and the release adds five new answer types: value reading, span finding, object localization boxes, ranged numerical estimates, and multi-label choices. Both models generate zero output tokens so only input is billed, at $0.042 and $0.02 per 1M tokens respectively, and are available only via the Liquid API with weights promised but undated. Calibration figures (0.8 points for d1.2-ultra across 13 benchmarks) come from Liquid AI's own testing.
Liquid AI d1.2decision modelscalibrated probabilities