Inception: Mercury

Inception: Mercury, the first diffusion large language model (dLLM), demonstrates exceptional speed and competitive pricing. It consistently ranks among the fastest models, achieving Infinityth percentile across 8 benchmarks, and offers among the most competitive pricing, also at the Infinityth percentile across 8 benchmarks. Mercury exhibits strong performance in several key areas. It achieves near-perfect accuracy in Ethics (100%), ranking in the top 3 for speed and being the most accurate model at its price point. In Hallucinations (Baseline), it shows high accuracy (98.0%), placing in the top 3 for speed and being the most accurate among models of comparable speed. General Knowledge (Baseline) also highlights its speed, achieving 97.5% accuracy and ranking #1 in speed. Coding (Baseline) is another strength, with 82.0% accuracy and a #1 speed ranking. However, the model shows significant weaknesses in Mathematics (Baseline) and Reasoning (Baseline), where it scored 0.0% accuracy in both. Email Classification (Baseline) and Instruction Following (Baseline) show moderate performance at 96.0% and 47.0% accuracy respectively, with the former being noted for its speed. Overall, Mercury excels in speed-sensitive applications and ethical considerations, but its current capabilities in complex mathematical and reasoning tasks are limited.