Can I deploy NVIDIA B300 or NVIDIA GH200 on Runcrate today?

NVIDIA B300 is not in stock on Runcrate right now; reserve a cluster, and once we've verified your company and your need we take your exact spec to GPU owners that fit. NVIDIA GH200 is not in stock on Runcrate right now; leave your email on its pricing page to register interest.

How much VRAM does each GPU have?

NVIDIA B300 has 288GB HBM3e. NVIDIA GH200 has 96GB HBM3.

B300 or GH200: 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 GH200: 1 GPU, 26 GB spare. Pick the B300: same GPU count, up to 2.0x the memory bandwidth.

GPU comparison

B300 vs GH200.

The B300 is ahead on memory (288 GB vs 96 GB) and speed (2.3x the FP16 throughput and up to 2.0x the memory bandwidth). The GH200 makes sense when your model fits in 96 GB and its hourly price is lower.

Reserve a cluster Deploy B200 instead

Rent either on Runcrate

B300Not in stock right now
Reserve a cluster B300 pricing
GH200Not in stock right now
Register interest GH200 pricing

Which is better for

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

JobB300GH200Pick

Hold a 70B model in 8-bit weights
70 GB of 8-bit weights
B300: 1 GPU, 218 GB spare
GH200: 1 GPU, 26 GB spare
B300. Same GPU count, up to 2.0x the memory bandwidth.
Serve an 8B model in 16-bit weights
16 GB of 16-bit weights
B300: 1 GPU, 272 GB spare
GH200: 1 GPU, 80 GB spare
B300. Same GPU count, up to 2.0x the memory bandwidth.
Fine-tune a 70B model with QLoRA
About 46 GB with 4-bit weights
B300: 1 GPU, 242 GB spare
GH200: 1 GPU, 50 GB spare
B300. Same GPU count, 2.3x the FP16 throughput.
Train on a full node
FP16 tensor throughput per node
B300: 8x, 18,000 TFLOPS, NVLink
GH200: 1x, 990 TFLOPS
B300. 8 GPUs per node, not 1.
Throughput per watt
FP16 TFLOPS per 100 W of TDP
B300: Varies by card
GH200: Superchip power, not a GPU TDP
Either. Not comparable.

Compare another pair

Blackwell

NVIDIA B300

Not in stock right now

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

Reserve a cluster Deploy B200 instead

Full pricing details

Hopper

NVIDIA GH200

Not in stock right now

Memory: 96 GB HBM3
FP16 tensor: 990 TFLOPS
Bandwidth: up to 4.0 TB/s

Full pricing details

At a glance

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

Memory

NVIDIA B300288 GB

NVIDIA GH20096 GB

Memory bandwidth

NVIDIA B300up to 8.0 TB/s

NVIDIA GH200up to 4.0 TB/s

FP16 tensor

NVIDIA B3002250 TFLOPS

NVIDIA GH200990 TFLOPS

FP32

NVIDIA B30075 TFLOPS

NVIDIA GH20067 TFLOPS

Full specs

The complete spec sheet.

Spec	NVIDIA B300	NVIDIA GH200
Architecture	Blackwell	Hopper
Memory	288 GB	96 GB
Memory type	HBM3e	HBM3
Memory bandwidth	up to 8.0 TB/s	up to 4.0 TB/s
FP32	75 TFLOPS	67 TFLOPS
FP16 tensor	2250 TFLOPS	990 TFLOPS
INT8	187.5 TOPS	1979 TOPS
Power	up to 1400 W per GPU (TDP)	up to 1000 W per superchip (CPU, GPU and memory)
Form factor	SXM	Superchip
NVLink	Yes	One GPU
GPUs per node	8	1
Hourly price (live)	Not in stock right now	Not in stock right now

Questions

B300 or GH200. What buyers ask first.

Other comparisons