Artificial Analysis releases Decision Index v2 with private datasets and calibration scoring
Overview
Artificial Analysis released Decision Index v2, which moves 45% of its weight onto private datasets and adds a Decision Score that rewards correct answers and calibrated confidence.
The company says the change better reflects real-world utility and guards against benchmark gaming.
In the post, Drex Max v1.1 from Nace AI, GLiDE v1.2 from Fastino Labs and Mammoth 27B from Baseten sit within one point at the top. Claude Haiku 5.5 (Max) scores 77 and GPT-6 Luna (Max) scores 76, but the post says each costs about 15 times more per 1M tasks than Drex Max v1.1.
Written by AI from the articles below · updated Nov 8, 8:25 PM ET
Check the sources:
Article timeline
The articles in this story. Times are ET.
- Artificial Analysis ArticlesOfficialArtificial Analysis releases Decision Index v2 with a calibration-based score
AIArtificial Analysis released Decision Index v2, which moves 45% of its weight onto private datasets and adds a Decision Score that rewards correct answers and calibrated confidence. Drex Max v1.1 from Nace AI, GLiDE v1.2 from Fastino Labs and Mammoth 27B from Baseten sit within one point at the top. The post says Claude Haiku 5.5 (Max) scores 77 and GPT-6 Luna (Max) scores 76 but cost about 15 times more per 1M tasks than Drex Max v1.1.
Heat Trend
Not enough continuous observations to show a trend yet.