workstation

Thuê Nvidia RTX PRO 5000.

Name: Nvidia RTX PRO 5000
Brand: Nvidia
SKU: nvidia-rtx-pro-5000

Blackwell 48GB VRAM 300W From $0.78/hr

Per hour

$0.78

Per day

$18.77

Per week

$131.41

Per month

$563

Provider spread 2 providers · up to 5% cheaper at the low end

Cheapest · $0.78/hr on Vast.ai Median $0.80/hr Most expensive · $0.82/hr on RunPod

Price history

Daily median across providers.

Where to rent it

All providers carrying this GPU.

Provider	$/hr	Range	Offers	30d trend	Updated
Vast.ai	$0.78/hr	$0.63/hr–$0.94/hr	10	↑ 7%	16 minutes	Launch ↗
RunPod	$0.82/hr	$0.82/hr–$0.82/hr	1	·	14 minutes	Launch ↗

Buy vs rent

Should you rent or own?

Your usage

Hours per day Electricity rate ($/kWh)

Rent on cloud

Per day

—

Per month

—

at $0.78/hr cheapest provider rate

Own on-prem

Electricity per day

—

300W TDP

AI models that fit See all 239 →

Run these on this GPU.

NVIDIA: Nemotron 3 Ultra 8× · int4
Reflection: Beam 1.0 501B A23B 8× · int4
Qwen: Qwen3 Coder 480B A35B 8× · int4

AI models

Models that run on this GPU.

Các mô hình mở trọng số chạy được trên card này để suy luận, với số card ít nhất và độ chính xác cần để tải mỗi mô hình. Tính hằng ngày từ kích thước mô hình và bộ nhớ của card. Hiển thị 60 mô hình lớn nhất trong 239 mô hình chạy được.

Renting for inference?

Pair these models with the cheapest provider on the rental table. See rental rates on the Overview tab →

FAQ

AI models on this GPU.

What AI models can I run on a Nvidia RTX PRO 5000?

The list above shows open-weights models whose memory need fits this card: on one card when the model uses at least a quarter of its 48 GB, or on 2, 4 or 8 of them when it is too large for one. Each row gives the fewest cards and the precision (FP16, FP8 or INT4) that load it.

What's the VRAM minimum to run a model on the Nvidia RTX PRO 5000?

We count the weights (parameters × bytes per weight: FP16 = 2, FP8 = 1, INT4 = 0.5), add the KV cache for an 8K-token context, and add 15% for the runtime. A model fits when that is within 90% of the cards' memory. A 70B model needs about 164 GB at FP16 and about 43 GB at INT4. A longer context needs more.

How does quantization (FP16 vs FP8 vs INT4) affect what fits?

Lower-precision quantization shrinks the memory footprint nearly linearly with the bit count. The trade-off is output quality: FP16 is the reference, FP8 is usually indistinguishable for most prompts, INT8 introduces small quality losses, INT4 is noticeably degraded on reasoning-heavy tasks but fine for chat. The badge on each row tells you which level the recommendation assumes.

Can I fine-tune on the Nvidia RTX PRO 5000 or only do inference?

The rows here are for inference. Full fine-tuning needs several times more memory than inference at the same model size, because gradients, optimizer state and activations all live in memory. LoRA and QLoRA cut that a lot. The fine-tune cost tool sizes a run for you.

How are these GPU counts worked out?

We compute them every day from the model's parameter count and the card's memory, with the formula above. They are estimates, not tests we ran. Tokens per second, where shown, is an estimate from the card's memory bandwidth on one card.

Thuê Nvidia RTX PRO 5000.

Daily median across providers.

All providers carrying this GPU.

Should you rent or own?

Run these on this GPU.

Models that run on this GPU.

NVIDIA: Nemotron 3 Ultra

Reflection: Beam 1.0 501B A23B

Qwen: Qwen3 Coder 480B A35B

Minimax M1 80K

MiniMax-Text-01

MiniMax: MiniMax M1

MiniMax: MiniMax M3

Baidu: ERNIE 4.5 VL 424B A47B

Nous: Hermes 3 405B Instruct

Nous: Hermes 4 405B

Llama 3.1 405B

Meta: Llama 4 Maverick

Arcee AI: Trinity Large Thinking

Qwen: Qwen3.5 397B A17B

Nex AGI: Nex-N2-Pro

Nex AGI: Nex-N2.5-Pro

Hunyuan-Large

GLM-4.7

GLM-4.6

GLM-4.5

Z.ai: GLM 5.3 Flash

Xiaomi: MiMo-V2.5

Xiaomi: MiMo-V2.6-Flash

Xiaomi: MiMo-V2-Flash

DeepSeek: DeepSeek V4 Flash Vision Exp

DeepSeek: DeepSeek V4 Flash 0731

Baidu: ERNIE 4.5 300B A47B

Tencent: Hy3 preview

Tencent: Hy3

DeepSeek: DeepSeek V4 Flash

Thinking Machines: Inkling Small

DeepSeek Coder V2 236B

Qwen: Qwen3 235B A22B Instruct 2507

Qwen 3 235B

Qwen: Qwen3 VL 235B A22B Thinking

Qwen: Qwen3 Max

Qwen: Qwen3 235B A22B Thinking 2507

Qwen: Qwen3 VL 235B A22B Instruct

MiniMax: MiniMax M2.5

MiniMax: MiniMax M2.7

MiniMax: MiniMax M2.1

MiniMax: MiniMax M2

MiniMax: MiniMax M3.1 Flash

StepFun: Step 3.7 Flash

Seed-1.8

StepFun: Step 3.5 Flash

Mixtral 8x22B

Mistral: Devstral 2 2512

Ling-3.0-flash

Ling 3.0 Flash Fin

Ling-3.0-flash-VL

Mistral Large 2

Qwen: Qwen3.5-122B-A10B

Mistral: Mistral Small 4

NVIDIA: Nemotron 3 Super

GPT-OSS 120B

Poolside: Laguna S 2.1

Llama 4 Scout (17Bx16E)

Meta: Llama 4 Scout

Z.ai: GLM 4.5V

AI models on this GPU.