LLM RAM Calculator → Hardware Guides

Cheapest A100 Rental Price — Comparison Table (2026)

A100 80GB SXM4 prices range from $1.89/hr (Vast.ai spot) to $3.20/hr (CoreWeave on-demand). This table compares every major provider. Always verify current pricing — this table is updated weekly.

Provider price comparison

ProviderSpot $/hrOn-demand $/hrNotes
Vast.ai$1.89$2.2Lowest spot price; variable availability
RunPod$2.09$2.49Reliable; community + secure cloud tiers
Lambda Labs$2.5On-demand only; good reliability
Hyperstack$2.2European regions available; NVLink clusters
CoreWeave$3.2Enterprise SLAs; highest reliability
Don't have the hardware? Rent it.

Verified GPU rental prices across all major providers. Start in minutes, pay by the hour.

Compare GPU prices →LLM RAM calculatorSelf-host vs API calculator

Frequently Asked Questions

What is the cheapest way to rent an A100?

Vast.ai spot instances offer the lowest A100 prices (~$1.89/hr) but availability fluctuates. For reliable access, RunPod community cloud ($2.09/hr spot) is a good balance. For production workloads needing SLAs, CoreWeave or Lambda are more appropriate despite higher prices.

A100 40GB vs 80GB — when does the extra VRAM matter?

An A100 80GB is worth the premium (~30% more) when running: (1) models larger than 20B at FP16, (2) large context windows with big KV-cache, (3) large batch inference for throughput. For models that fit in 40GB (< 20B at FP16), the 40GB is more cost-efficient.

Data verified 2026-08-20. GPU rental prices change frequently — verify before purchasing. See methodology.