Against the backdrop of rapid advancements in large model technology, token services have emerged as a pivotal component in its deployment. The report jointly released by relevant institutions highlights that, in response to users' challenges in assessing token service performance and selecting appropriate models, the report conducts a comprehensive analysis of large model token services both domestically and globally. This analysis is based on monitoring data collected from January to August 2026, covering multiple dimensions. According to the report, domestic token services are witnessing a simultaneous uptick in both demand and pricing. The standardized average price in 2026 has nearly doubled compared to the second half of 2025, accompanied by a significant surge in token consumption. The average daily consumption in June 2026 reached approximately 500 trillion tokens, nearly quadrupling that of December 2025. Nevertheless, the average token price overseas continues to outpace domestic prices by a considerable margin. Regarding performance, the number of tokens generated per second by domestic models has doubled compared to 2025, yet they still fall short of the overall performance of typical foreign models. Both domestic and foreign models have experienced an increase in first-token latency, a trend attributed to the concurrent rise in model parameter count, context length, and invocation frequency, rather than a mere decline in performance. Presently, long-context models have gained substantial traction, with models featuring 200,000 tokens or more constituting 95.7% of the total. The report also compares the performance of four open-source models across various third-party platforms, revealing consistent performance improvements with each new model version. However, notable performance disparities exist for the same model across different platforms, with the maximum TPS (Transactions Per Second) difference approaching double. This is primarily attributable to variations in inference acceleration techniques, optimization strategies, and cost-control measures across platforms.