Amazon Bedrock 在泰國、馬來西亞、新加坡、印尼與台灣推出 Claude 模型全球跨區域推論

🔑 增強重點摘要

• •Global CRIS provides approximately 10% savings on input and output token pricing compared to geographic cross-Region inference profiles.[6]
• •Global CRIS routes requests across more than 20 AWS commercial Regions for higher availability, supporting on-demand inference, batch inference, agents, model evaluation, prompt management, and prompt flows.[1][6]
• •Claude Opus 4.6 excels in demanding enterprise workloads, Sonnet 4.6 offers balanced performance approaching Opus intelligence at lower cost for coding and knowledge work, and Haiku 4.5 enables cost-efficient high-volume operations.[1][5]

🛠️ 技術深入

• •To invoke Global CRIS, specify the global inference profile ID such as global.anthropic.claude-sonnet-4-5-20250929-v1:0 in API calls instead of region-specific model IDs, with IAM permissions for destination Regions.[6]
• •Global CRIS optimizes resource utilization by dynamically routing requests worldwide beyond geographic boundaries, unlike GEO CRIS which keeps processing within specific areas like Japan or Australia.[2][6]
• •Supported in source Regions including ap-southeast-1 (Singapore) and ap-southeast-2, with destination routing to global AWS infrastructure for quota flexibility and handling traffic spikes.[6][7]

🔮 前景展望AI analysis grounded in cited sources

⏳ 時間線