Question 1

What AI models can run on A40?

Accepted Answer

The A40 (48 GB VRAM) can run models including gemma-3-12b-it, gemma-3-4b-it, gemma-3-4b-pt, Llama-4-Scout-17B-16E, gemma-3-12b-pt and more — 10 compatible models indexed so far. Compatible models are determined by whether the model's memory footprint fits within the GPU's 48 GB VRAM, potentially using quantization (Q8, Q6, or Q4) to reduce memory usage.

Question 2

What is the cheapest cloud option for A40?

Accepted Answer

The cheapest on-demand option for the A40 is RunPod at $0.44/hr (1x_A40_SECURE). Spot/preemptible pricing is available at $0.20/hr. We track pricing across 8 providers to help you find the best deal.

Question 3

Is A40 good for fine-tuning?

Accepted Answer

Yes, the A40 is well-suited for fine-tuning. With 48 GB of VRAM, it can handle LoRA/QLoRA fine-tuning of models up to ~30B parameters and full fine-tuning of smaller models (up to ~7B parameters). For larger models, consider multi-GPU setups.

Question 4

How much VRAM does A40 have?

Accepted Answer

The A40 has 48 GB of VRAM (GDDR6). It also offers 696 GB/s memory bandwidth, 149.7 TFLOPS of FP16 compute.

Question 5

What quantization formats work on A40?

Accepted Answer

The A40 supports the following quantization formats: FP16 (full precision), Q8 (8-bit), Q6 (6-bit), Q4 (4-bit). FP16 works for models up to ~7B parameters. Q4 quantization lets models up to ~163B parameters (by VRAM estimate) fit on the A40.

A40

Specifications

Cloud pricing

Compatible models

Compare this GPU

Frequently asked questions

VRAM	48 GB
Memory type	GDDR6
Bandwidth	696 GB/s
FP16 TFLOPS	149.7
FP8 support	No

A40