聚焦中国互联网:人工智能模型架构的战略意义。token price of RMB4/MTok (c. $0.60/MTok). Exhibit 6 and Exhibit 7 show OpenRouter data on cache hit rates across various providers for Minimax M2.5 and Z.ai GLM5. Ignoring outliers which report zero, the former currently has a c. 70% median cache hit rate, compared with the latter on c. 40%.
聚焦中国互联网:人工智能模型架构的战略意义。token price of RMB4/MTok (c. $0.60/MTok). Exhibit 6 and Exhibit 7 show OpenRouter data on cache hit rates across various providers for Minimax M2.5 and Z.ai GLM5. Ignoring outliers which report zero, the former currently has a c. 70% median cache hit rate, compared with the latter on c. 40%.
聚焦中国互联网:人工智能模型架构的战略意义。token price of RMB4/MTok (c. $0.60/MTok). Exhibit 6 and Exhibit 7 show OpenRouter data on cache hit rates across various providers for Minimax M2.5 and Z.ai GLM5. Ignoring outliers which report zero, the former currently has a c. 70% median cache hit rate, compared with the latter on c. 40%.
覆盖 2026 年,全文共 16 页。
本报告为付费报告,可在资料宝站内下单后获取完整内容。
适合战略规划、产品经理、行业研究员等读者参考。
重点分析了人工智能、中国互联网、人工智能模型架构的战略意义等议题。