Motif Technologies: Motif 3

Motif 3 from Motif Technologies is a high-capacity Mixture-of-Experts language model featuring an expansive 262,144-token context window. Overall, the model delivers competitive response times, placing in the 54th percentile for speed, while maintaining cost-effective pricing that ranks in the 61st percentile across evaluated tasks. Notably, Motif 3 exhibits exceptional operational reliability, achieving an outstanding 99% success rate with virtually zero technical failures. Across benchmark categories, Motif 3 displays distinct strengths and moderate limitations. Its strongest showing is in mathematics, where it achieved a 92.0% accuracy rate alongside lightning-fast execution speed in the 96th percentile. The model also performs well in complex instruction following, attaining 74.0% accuracy (67th percentile), demonstrating solid fidelity to multi-layered directives. Conversely, its primary weaknesses lie in general deductive reasoning and coding; it registered 70.0% accuracy (38th percentile) in multi-step reasoning and 86.0% accuracy (44th percentile) on coding evaluations, falling slightly below industry medians. Supported by strong multilingual capabilities in English and Korean, Motif 3 represents a dependable, budget-friendly solution for tasks demanding rigorous math comprehension, large context ingestion, and reliable directive execution.