Better AI spend visibility lands in AWS Cost Explorer and Budgets

Amazon Bedrock product attributes are now available in AWS Cost Explorer, AWS Budgets, and AWS Cost Management Dashboards. This update lets you analyze Bedrock spend by model, model provider, inference type, and feature, so you can see which workloads are actually driving your AI bill.

You can group and filter costs by attributes like model, provider, input tokens, output tokens, on-demand inference, and reranker. In Budgets, you can also filter budgets by the same attributes, which gives finance and platform teams a clearer way to track specific Bedrock usage.

Claude Haiku 5.5 brings a lower-cost option for AI workloads

Claude Haiku 5.5 is now available on AWS. AWS says it’s the fastest and most efficient model in the Claude 5.5 family, which makes it a strong fit for cost-sensitive workloads.

Anthropic says it costs around 75% less than Claude Haiku 4.5 for most tasks. For teams running lots of AI inference, that can translate into noticeably better cost efficiency without changing the overall workflow.

Lambda Web Functions makes HTTP workloads easier to cost-track

AWS Lambda announces Lambda Web Functions. AWS describes it as a new serverless compute primitive for HTTP workloads, and the billing model focuses on active CPU time and memory instead of idle CPU time.

That shift is useful for web apps, APIs, streaming endpoints, and other HTTP-based workloads. It gives teams a way to pay more directly for actual activity, which can improve cost efficiency for traffic patterns that don’t run constantly.

S3 Express One Zone expands high-performance storage options

Amazon S3 Express One Zone is now available in 4 additional AWS Regions. This expands access to a storage class built for high-performance, latency-sensitive workloads.

AWS says it can deliver up to 10x faster data access and up to 80% lower request costs than S3 Standard. That can make a real difference for teams that care about both performance and request-heavy storage economics.

ECS adds a better scaling metric and drops a metric charge

Amazon ECS adds LiveTaskCount and removes the charge for RunningTaskCount metric. The new LiveTaskCount metric is designed to better reflect actively running tasks for autoscaling and capacity planning.

That makes it easier to base scaling and planning decisions on a metric that better matches what’s actually running. At the same time, AWS removed the charge for the existing RunningTaskCount metric in Container Insights, which lowers observability cost for ECS users.

C8gb instances are now available in more regions

Amazon EC2 C8gb instances now generally available in additional regions. These Graviton4-powered instances are now available across several major regions, including US East, US West, Europe, and Africa.

AWS says the new instances offer better performance and higher EBS throughput. That can help storage-intensive applications run more efficiently and may improve total cost of ownership.

AWS Config expands coverage across more resource types

AWS Config now supports 77 new resource types. The added support includes resources across services such as EC2 and S3.

Broader coverage improves governance, inventory accuracy, and cost and compliance visibility across larger environments. That helps teams get a more complete picture of what’s in use and where risk or waste might be hiding.

ElastiCache for Valkey gets better operational insight

Amazon ElastiCache for Valkey introduces ElastiCache Insights. The new CloudWatch-based dashboard includes health, memory, network, latency, and throughput metrics.

That added observability can help teams right-size cache clusters and spot inefficient or overprovisioned deployments sooner. In practice, that makes it easier to see whether a cache is actually earning its keep.