t4g.xlargeAWS Pricing, Specs & Benchmarks

The t4g family runs on AWS Graviton2 — ARM-based processors that AWS designed in-house. Compared to equivalently sized Intel/AMD burstable instances (t3 and t3a), t4g instances are typically 20–40% cheaper per hour while delivering comparable single-thread performance for general-purpose workloads. t4g.xlarge at $0.1344/hr On-Demand reflects that Graviton pricing discount: a t3 or t3a equivalent at the same vCPU/memory ratio usually runs noticeably higher.

Like other "t" instances, t4g uses a burstable CPU model: you get a baseline of CPU performance and earn credits when idle, which the instance spends on bursts above the baseline. This makes t4g a strong fit for web servers, small databases, dev/test environments, microservices, and CI build agents — workloads that are bursty rather than consistently CPU-bound. For sustained-CPU workloads (continuous data processing, batch encoding, heavy compilation), look at c7g or m7g instead; those use Graviton3 without the burstable cap.

The main gotcha is architecture: t4g runs aarch64 (ARM64), not x86_64. Most modern Linux distros, Node.js, Python, Go, Rust, Java, and PHP have first-class ARM64 builds, and Docker images are typically multi-arch. But you should verify any compiled extensions (Node native modules, Python wheels with C extensions) and any third-party binaries you depend on. If you're migrating from a t3 or t3a, a one-line change in your AMI ID is rarely the whole story.

t4g.xlarge provides 4 vCPUs and 16GB of RAM. On Spot, the price drops to $0.0523/hr — roughly 61% off On-Demand — but Spot capacity for Graviton instances can be tighter than equivalent Intel SKUs in some regions, so plan for interruption handling. See the regional pricing table below for the cheapest region today, or compare against the rest of the t4g family further down.