EC2 Capacity Blocks for ML GPU prices increased ~15% across regions. p5e.48xlarge went from $34.61 to $39.80/hr (US East) and $43.26 to $49.75/hr (US West N. California)
Forge platform moved from free to consumption-based billing for compute and storage
Volume snapshots now billed — previously free automatic snapshots now cost $0.08/GB-month
Pro plan moved to credit-based model ($20/mo credit pool), Hobby plan metrics renamed and restructured — bandwidth→Fast Data Transfer, serverless split into Active CPU + Provisioned Memory, Blob Storage and Image Transformations added
20% price cut on Basic Droplets (cheapest now $4/mo from $5/mo). Per-second billing introduced (minimum 60 seconds or $0.01). Functions free tier increased to 25,000 GiB-seconds/mo