Recent evaluations of large language models (LLMs) highlight the competitive landscape among leading technologies. Sonnet 3.5 has been recognized as the best coding LLM, outperforming others like Qwen and o1-mini, which are noted as competitive but not at the same level. Benchmark scores reveal that o1-preview achieved 71.4%, while o1-mini scored 52.6%, and Sonnet 3.5 lagged behind with 33.4%. Additionally, a new model, QwQ, which is an open-source 32B model, has shown impressive reasoning capabilities, ranking second overall, just behind the o1 line. In terms of pricing and performance, Google’s offerings, including Gemini-Exp models, are noted for their competitive pricing, with Google standing out in price/performance comparisons. Overall, the evaluations suggest a significant gap in performance metrics among the top models, with Chinese open-weight LLMs also making notable strides in mathematical capabilities.