Pricing
Cloud GPU Pricing
Current list-price hourly rates for GPU instances on AWS, Azure, and GCP. Refreshed daily from each cloud's public pricing API.
Refreshed — · —
View pricing →
Soterra Labs Anvil
Reference data for AI infrastructure decisions.
Free public reference content, refreshed automatically from public APIs and benchmark publications. Every value on every page traces back to its source. No editorial commentary, no human-edited numbers.
Benchmarks
MLPerf Inference Results
Latest schema-audited MLPerf Inference Datacenter results, filtered to common workloads. Every row links to its official MLCommons submission.
Round v5.1 · Ingested —
View results →
How Anvil works
Each Anvil page is rebuilt on a fixed cadence by a fully automated pipeline: scheduled fetcher pulls public data, validates plausibility against bounded ranges, commits to an audit-traceable database, and rebuilds the page. No human edits values; humans only update name-mapping configuration when a new instance type or accelerator appears in upstream sources. When a fetch fails, the previous good page stays live and a banner indicates the data is stale.