Tencent: Hunyuan A13B Instruct

Tencent's Hunyuan A13B Instruct model, an 80B total parameter MoE model with 13B active parameters, demonstrates strong performance across several key metrics. It consistently ranks among the fastest models and offers highly competitive pricing, positioning it as an efficient and cost-effective solution. The model exhibits high reliability with a 95% success rate across benchmarks, indicating consistent operational stability. In terms of capabilities, Hunyuan A13B excels in Reasoning, achieving 96.0% accuracy, placing it in the 86th percentile. Its General Knowledge is also robust at 97.5% accuracy, though this places it in the 49th percentile relative to other models. Coding performance is solid at 90.0% accuracy (66th percentile). However, the model shows significant weaknesses in Instruction Following, with 0.0% accuracy in both tests, suggesting a critical area for improvement. Mathematics performance is moderate at 59.0% accuracy (22nd percentile), and Email Classification is 91.0% accurate but ranks in the 16th percentile. Its support for Chain-of-Thought reasoning and inference efficiency features like GQA and quantization are notable technical advantages.