Qwen: Qwen3 30B A3B Instruct 2507

Qwen3-30B-A3B-Instruct-2507 demonstrates a strong overall performance profile, positioning itself as a highly reliable and competitive model. It consistently performs among the faster models, ranking in the 63rd percentile for speed, and offers competitive pricing, placing in the 57th percentile. A standout feature is its exceptional reliability, achieving a perfect 100% success rate across all benchmarks, indicating minimal technical failures. In terms of specific performance, the model excels in Coding (Baseline) with 93.0% accuracy, notably being the most accurate model at its price point. It achieved perfect accuracy (100.0%) in the Ethics (Baseline) benchmark, further highlighting its strong ethical reasoning capabilities and making it the most accurate model at its price and speed. While its General Knowledge (97.0%) and Reasoning (74.0%) scores are solid, its Instruction Following (59.8%) and Email Classification (94.0%) performance, though respectable, are not as high in their respective percentiles. The model's design for high-quality instruction following and agentic tool use is evident, and its competitive performance across reasoning, coding, and alignment benchmarks makes it a versatile choice.