A40 rental price: from $0.59/hr today

A40 offers: 2 configurations from 2 providers

On-demand and secure-cloud rates are capacity the provider runs, with no reservation. Spot and community capacity can be interrupted. A marketplace host is one independent owner's server listed on a marketplace. Prices are per GPU per hour, before storage and data transfer. Every row shows the date it was checked.

A40 price history and range

A40 prices by provider and billing type

Each provider's cheapest rate in each billing type, per GPU-hour. The provider name opens its A40 page with node sizes, fees and its own history.

A40 hardware specs

Manufacturer datasheet figures for the A40. Source: images.nvidia.com.

LLM memory estimates for this GPU

These models have an estimated inference footprint within 48 GB at one or more precisions. The scenario uses 4,096 tokens, batch 1, a 16-bit KV cache and 20% runtime headroom. A memory fit does not establish speed or software compatibility.

Models that need more than one A40

• Mixtral 8x22B v0.1 at 16-bit needs about 339 GB with the same context and headroom, so at least 8 cards. No provider we track publishes a A40 node of that size today. At 8 or 4 bit it needs less; the model page recalculates.
• Qwen2.5 72B at 16-bit needs about 176 GB with the same context and headroom, so at least 4 cards. No provider we track publishes a A40 node of that size today. At 8 or 4 bit it needs less; the model page recalculates.
• DeepSeek R1 Distill Llama 70B at 16-bit needs about 171 GB with the same context and headroom, so at least 4 cards. No provider we track publishes a A40 node of that size today. At 8 or 4 bit it needs less; the model page recalculates.

Compare model memory requirements or change context and memory assumptions on a model's page.

When to choose the A40, and what to pick instead

Start from memory: your model, its context and your batch must fit in 48 GB (the model-fit table above estimates it). Then the A40 is the right card unless one of these fits your job better.

• Pick the A100 (80 GB, from $1.09/hr) when the model or its cache does not fit in 48 GB.
• Pick the L40 (48 GB, from $0.79/hr) when finishing sooner is worth paying more: it has 1.2 times the A40's 16-bit compute.

Rent or buy the A40

We have no confirmed A40 purchase price from the stores we check; renting starts at $0.59 per GPU-hour. A seller quote entered in the calculator gives the break-even.

Compare A40 purchase prices and quote availability, then enter your usage and electricity costs in the rent-versus-buy calculator. Purchase listings have their own variants, stock status and update dates. Compare two GPUs.

A40 questions

How much does it cost to rent an A40?

The cheapest A40 rental we have is $0.59 per GPU-hour on RunPod (secure cloud), checked 13 October 2026.

How much VRAM does an A40 have?

48 GB.

Where is the A40 cheapest right now?

RunPod has the cheapest checked on-demand rate today at $0.59 per GPU-hour.

GPUs near the A40

• RTX 6000 Ada48 GB$0.75/hrcheapest on-demand
• L4048 GB$0.79/hrcheapest on-demand
• L40S48 GB$0.96/hrcheapest on-demand
• RTX A600048 GB$0.35/hrcheapest on-demand
• A10040-80 GB$0.89/hrcheapest on-demand

Related

All 48 GB+ GPUs · GPUs under $1/hr · Price movers this week · RunPod review · L40S vs A40 · A100 vs A40 · A10 vs A40