What's the price difference between NVIDIA H200 NVL and NVIDIA RTX 4090?

On Runcrate today, NVIDIA H200 NVL starts at $4.63/hr and NVIDIA RTX 4090 starts at $0.68/hr; NVIDIA RTX 4090 is cheaper by 85%.

Can I deploy NVIDIA H200 NVL or NVIDIA RTX 4090 on Runcrate today?

NVIDIA H200 NVL is on demand in 1 region from $4.63/hr, billed by the second on most GPUs. NVIDIA RTX 4090 is on demand in 1 region from $0.68/hr, billed by the second on most GPUs.

How much VRAM does each GPU have?

NVIDIA H200 NVL has 141GB HBM3e. NVIDIA RTX 4090 has 24GB GDDR6X.

H200 NVL or RTX 4090: which holds a 70B model in 8-bit weights?

It needs 70 GB of 8-bit weights. On the H200 NVL: 1 GPU, 71 GB spare. On the RTX 4090: 4 GPUs, more than one 2x node. Pick the H200 NVL: fits on one GPU.

GPU comparison

H200 NVL vs RTX 4090.

The H200 NVL is ahead on memory (141 GB vs 24 GB) and speed (5.1x the FP16 throughput and 4.8x the memory bandwidth). The RTX 4090 makes sense when your model fits in 24 GB and its hourly price is lower.

Launch RTX 4090 Reserve a cluster

Rent either on Runcrate

H200 NVLfrom $4.63/hr
$0.033 per GB/hr$0.55 per 100 TFLOPS/hr
Launch H200 NVL H200 NVL pricing
RTX 4090Cheaperfrom $0.68/hr
$0.028 per GB/hr$0.41 per 100 TFLOPS/hr
Launch RTX 4090 RTX 4090 pricing

Live hourly price, then per GB of memory and per 100 FP16 TFLOPS.

Which is better for

Which to pick for each job. Memory fit and throughput, from the datasheets.

JobH200 NVLRTX 4090Pick

Hold a 70B model in 8-bit weights
70 GB of 8-bit weights
H200 NVL: 1 GPU, 71 GB spare
RTX 4090: 4 GPUs, more than one 2x node
H200 NVL. Fits on one GPU.
Serve an 8B model in 16-bit weights
16 GB of 16-bit weights
H200 NVL: 1 GPU, 125 GB spare
RTX 4090: 1 GPU, 8 GB spare
H200 NVL. Same GPU count, 4.8x the memory bandwidth.
Fine-tune a 70B model with QLoRA
About 46 GB with 4-bit weights
H200 NVL: 1 GPU, 95 GB spare
RTX 4090: 2 GPUs (48 GB)
H200 NVL. Fits on one GPU.
Train on a full node
FP16 tensor throughput per node
H200 NVL: 8x, 6,680 TFLOPS, NVLink in groups of 4
RTX 4090: 2x, 330 TFLOPS
H200 NVL. 8 GPUs per node, not 2.
Throughput per watt
FP16 TFLOPS per 100 W of TDP
H200 NVL: 139 TFLOPS per 100 W
RTX 4090: 37 TFLOPS per 100 W
H200 NVL. 3.8x the FP16 per watt.

Compare another pair

Hopper

NVIDIA H200 NVL

from$4.63/hr

Available now · 1 region

Memory: 141 GB HBM3e
FP16 tensor: 835 TFLOPS
Bandwidth: 4.8 TB/s

Launch H200 NVL Reserve a cluster

Full pricing details

Ada Lovelace

NVIDIA RTX 4090

from$0.68/hrCheaper

Available now · 1 region

Memory: 24 GB GDDR6X
FP16 tensor: 165.2 TFLOPS
Bandwidth: 1.0 TB/s

Launch RTX 4090 Reserve a cluster

Full pricing details

At a glance

Spec by spec. The longer bar is the better number.

Memory

NVIDIA H200 NVL141 GB

NVIDIA RTX 409024 GB

Memory bandwidth

NVIDIA H200 NVL4.8 TB/s

NVIDIA RTX 40901.0 TB/s

FP16 tensor

NVIDIA H200 NVL835 TFLOPS

NVIDIA RTX 4090165.2 TFLOPS

Hourly price, from (live)

NVIDIA H200 NVL$4.63/hr

NVIDIA RTX 4090$0.68/hr

Full specs

The complete spec sheet.

Spec	NVIDIA H200 NVL	NVIDIA RTX 4090
Architecture	Hopper	Ada Lovelace
Memory	141 GB	24 GB
Memory type	HBM3e	GDDR6X
Memory bandwidth	4.8 TB/s	1.0 TB/s
FP32	60 TFLOPS	82.6 TFLOPS
FP16 tensor	835 TFLOPS	165.2 TFLOPS
INT8	1670 TOPS	661 TOPS
TDP	600 W	450 W
Form factor	PCIe	Consumer
NVLink	In groups of 4	No
GPUs per node	8	2
Hourly price (live)	from $4.63/hr	from $0.68/hr

Questions

H200 NVL or RTX 4090. What buyers ask first.

Other comparisons