Skip to content

VRAM needed and GPU rental cost, by model

How much GPU memory popular open models need for inference and fine-tuning, and the cheapest public on-demand rental that fits. Parameter counts come from each model card. Try the calculator.

16 models

Pick a model

Estimates use parameters times bytes per parameter plus a 20% allowance for the KV cache and activations. Prices come from our GPU price pages.

Llama 3.1 8B8B parametersAbout 19 GB at FP16, 10 GB at FP8, 5 GB at INT4. Cheapest FP8 rental from $0.97/hr.Llama 3.1 8B guide Llama 3.1 70B70.6B parametersAbout 169 GB at FP16, 85 GB at FP8, 42 GB at INT4. Cheapest FP8 rental from $1.94/hr.Llama 3.1 70B guide Llama 3.3 70B70.6B parametersAbout 169 GB at FP16, 85 GB at FP8, 42 GB at INT4. Cheapest FP8 rental from $1.94/hr.Llama 3.3 70B guide Llama 3.1 405B405.9B parametersAbout 974 GB at FP16, 487 GB at FP8, 244 GB at INT4. Cheapest FP8 rental from $7.63/hr.Llama 3.1 405B guide Llama 4 Scout108.6B parametersAbout 261 GB at FP16, 130 GB at FP8, 65 GB at INT4. Cheapest FP8 rental from $2.18/hr.Llama 4 Scout guide Mistral 7B7.2B parametersAbout 17 GB at FP16, 9 GB at FP8, 4 GB at INT4. Cheapest FP8 rental from $0.97/hr.Mistral 7B guide Mixtral 8x7B46.7B parametersAbout 112 GB at FP16, 56 GB at FP8, 28 GB at INT4. Cheapest FP8 rental from $1.09/hr.Mixtral 8x7B guide Qwen2.5 7B7.6B parametersAbout 18 GB at FP16, 9 GB at FP8, 5 GB at INT4. Cheapest FP8 rental from $0.97/hr.Qwen2.5 7B guide Qwen2.5 32B32.8B parametersAbout 79 GB at FP16, 39 GB at FP8, 20 GB at INT4. Cheapest FP8 rental from $0.97/hr.Qwen2.5 32B guide Qwen2.5 72B72.7B parametersAbout 174 GB at FP16, 87 GB at FP8, 44 GB at INT4. Cheapest FP8 rental from $1.94/hr.Qwen2.5 72B guide Qwen3 32B32.8B parametersAbout 79 GB at FP16, 39 GB at FP8, 20 GB at INT4. Cheapest FP8 rental from $0.97/hr.Qwen3 32B guide Qwen3 30B-A3B30.5B parametersAbout 73 GB at FP16, 37 GB at FP8, 18 GB at INT4. Cheapest FP8 rental from $0.97/hr.Qwen3 30B-A3B guide Gemma 2 27B27.2B parametersAbout 65 GB at FP16, 33 GB at FP8, 16 GB at INT4. Cheapest FP8 rental from $0.97/hr.Gemma 2 27B guide DeepSeek V3684.5B parametersAbout 1643 GB at FP16, 821 GB at FP8, 411 GB at INT4. Cheapest FP8 rental from $17.25/hr.DeepSeek V3 guide gpt-oss-120b116.8B parametersAbout 280 GB at FP16, 140 GB at FP8, 70 GB at INT4. Cheapest FP8 rental from $2.18/hr.gpt-oss-120b guide gpt-oss-20b20.9B parametersAbout 50 GB at FP16, 25 GB at FP8, 13 GB at INT4. Cheapest FP8 rental from $0.97/hr.gpt-oss-20b guide

Need this capacity?

Tell us the GPU, count, term, region and budget. We reply with the public prices that match and the checks to run before you pay. Free, and we do not sell compute.

Send a GPU request