Qwen: Qwen3 30B A3B

Qwen3 30B A3B, the latest Qwen model, demonstrates a strong overall performance profile, particularly excelling in reliability and specific cognitive tasks. While its speed performance is moderate, ranking in the 20th percentile across benchmarks, it offers competitive pricing, placing in the 51st percentile. A standout feature is its exceptional reliability, achieving a 100% success rate across all 8 benchmarks, indicating minimal technical failures and consistent response generation. The model exhibits significant strengths in knowledge-based and ethical reasoning tasks, achieving perfect 100% accuracy in both General Knowledge and Ethics benchmarks. It also performs very well in Coding (94.0% accuracy, 89th percentile) and Reasoning (96.0% accuracy, 85th percentile), showcasing its advanced capabilities in complex problem-solving. Its ability to handle multilingual support and advanced agent tasks, as highlighted in its description, is reflected in these strong reasoning scores. Hallucination rates are low at 96.0% accuracy, indicating a good understanding of uncertainty. While Mathematics performance is solid at 93.0% accuracy (73rd percentile), Instruction Following is a relative weakness at 60.8% accuracy (58th percentile), suggesting room for improvement in adhering to highly complex, multi-step instructions. The model's unique architecture, combining dense and MoE components, appears to contribute to its high accuracy in critical areas.