What's the price difference between NVIDIA B300 and NVIDIA RTX 4090?

On Runcrate today, NVIDIA B300 starts at $10.40/hr and NVIDIA RTX 4090 starts at $0.57/hr per GPU on an 8-GPU machine; NVIDIA RTX 4090 is cheaper by 95%.

Can I deploy NVIDIA B300 or NVIDIA RTX 4090 on Runcrate today?

NVIDIA B300 is on demand in 1 region from $10.40/hr, billed by the second on most GPUs. NVIDIA RTX 4090 is on demand in 4 regions from $0.57/hr per GPU on an 8-GPU machine, billed by the second on most GPUs.

How much VRAM does each GPU have?

NVIDIA B300 has 288GB HBM3e. NVIDIA RTX 4090 has 24GB GDDR6X.

B300 or RTX 4090: which holds a 70B model in 8-bit weights?

It needs 70 GB of 8-bit weights. On the B300: 1 GPU, 218 GB spare. On the RTX 4090: 4 GPUs, more than one 2x node. Pick the B300: fits on one GPU.

GPU comparison

B300 vs RTX 4090.

The B300 is ahead on memory (288 GB vs 24 GB) and speed (13.6x the FP16 throughput and up to 7.9x 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

B300from $10.40/hr
$0.036 per GB/hr$0.46 per 100 TFLOPS/hr
Launch B300
RTX 4090Cheaperfrom $0.57/hr per GPU
on an 8-GPU machine
$0.024 per GB/hr$0.35 per 100 TFLOPS/hr
Launch RTX 4090

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.

JobB300RTX 4090Pick

Hold a 70B model in 8-bit weights
70 GB of 8-bit weights
B300: 1 GPU, 218 GB spare
RTX 4090: 4 GPUs, more than one 2x node
B300. Fits on one GPU.
Serve an 8B model in 16-bit weights
16 GB of 16-bit weights
B300: 1 GPU, 272 GB spare
RTX 4090: 1 GPU, 8 GB spare
B300. Same GPU count, up to 7.9x the memory bandwidth.
Fine-tune a 70B model with QLoRA
About 46 GB with 4-bit weights
B300: 1 GPU, 242 GB spare
RTX 4090: 2 GPUs (48 GB)
B300. Fits on one GPU.
Train on a full node
FP16 tensor throughput per node
B300: 8x, 18,000 TFLOPS, NVLink
RTX 4090: 2x, 330 TFLOPS
B300. 8 GPUs per node, not 2.
Throughput per watt
FP16 TFLOPS per 100 W of TDP
B300: Varies by card
RTX 4090: 37 TFLOPS per 100 W
Either. Varies by card.

Compare another pair

Blackwell

NVIDIA B300

from$10.40/hr

Available now · 1 region

Memory: 288 GB HBM3e
FP16 tensor: 2250 TFLOPS
Bandwidth: up to 8.0 TB/s

Launch B300 Reserve a cluster

Full pricing details

Ada Lovelace

NVIDIA RTX 4090

from$0.57/hr per GPUCheaper

on an 8-GPU machine

Available now · 4 regions

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 B300288 GB

NVIDIA RTX 409024 GB

Memory bandwidth

NVIDIA B300up to 8.0 TB/s

NVIDIA RTX 40901.0 TB/s

FP16 tensor

NVIDIA B3002250 TFLOPS

NVIDIA RTX 4090165.2 TFLOPS

Hourly price, from (live)

NVIDIA B300$10.40/hr

NVIDIA RTX 4090$0.57/hr

Full specs

The complete spec sheet.

Spec	NVIDIA B300	NVIDIA RTX 4090
Architecture	Blackwell	Ada Lovelace
Memory	288 GB	24 GB
Memory type	HBM3e	GDDR6X
Memory bandwidth	up to 8.0 TB/s	1.0 TB/s
FP32	75 TFLOPS	82.6 TFLOPS
FP16 tensor	2250 TFLOPS	165.2 TFLOPS
INT8	187.5 TOPS	661 TOPS
TDP	up to 1400 W	450 W
Form factor	SXM	Consumer
NVLink	Yes	No
GPUs per node	8	2
Hourly price (live)	from $10.40/hr	from $0.57/hr per GPU on an 8-GPU machine

Questions

B300 or RTX 4090. What buyers ask first.

Other comparisons