NVIDIA B300 SXM 288GB vs NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip 252GB

The GB300 delivers 11% more FP8 throughput than the B300 (5,000 vs 4,500 TFLOPS dense). The B300 counters with 36 GB more memory (288 GB vs 252 GB).

B300: Blackwell Ultra, 2025 GB300: Blackwell Ultra, 2025
FP8 dense lead
+11%
GB300: 5,000 vs 4,500 TFLOPS
VRAM
288 vs 252 GB
36 GB more for the B300
Bandwidth
8.0 TB/s vs 7.1 TB/s
B300 moves data faster

Specifications

B300
GB300
Architecture
Blackwell UltraBlackwell Ultra
Launch Year
20252025
Form Factor
SXMSuperchip
VRAM
288 GB252 GB
Memory Bandwidth
8.0 TB/s7.1 TB/s
TDP
1.4 kW
Process Node
4nm4nm

Performance (TFLOPS)

B300
GB300
FP64
1.3 TFlops 1.3 TFlops
FP32
75.0 TFlops 80.0 TFlops
FP16
2250.0 TFlops
4500.0 TFLOPS sparse
2500.0 TFlops
5000.0 TFLOPS sparse
BF16
2250.0 TFlops
4500.0 TFLOPS sparse
2500.0 TFlops
5000.0 TFLOPS sparse
FP8
4500.0 TFlops
9000.0 TFLOPS sparse
5000.0 TFlops
10000.0 TFLOPS sparse
INT8
187.5 TFlops
375.0 TOPS sparse
165.0 TFlops
330.0 TOPS sparse

FLOPS by Precision

What actually differs

Both GPUs launched in 2025: the B300 on Blackwell Ultra and the GB300 on Blackwell Ultra.

Dense throughput for the B300 against the GB300: FP64 1.3 vs 1.3 TFLOPS, FP32 75 vs 80 TFLOPS, FP16 2,250 vs 2,500 TFLOPS, FP8 4,500 vs 5,000 TFLOPS. Sparse figures, where the vendor publishes them, appear under each dense number in the performance table.

Memory is 288 GB against 252 GB, fed at 8.0 TB/s versus 7.1 TB/s. More memory per GPU means larger models fit before you have to shard across cards, which often matters more than raw TFLOPS for inference.

For LLM training and inference, weight the FP8 and FP16 rows and memory capacity most heavily. For scientific and HPC workloads, the FP64 row is the one to read.

Frequently asked questions

Is the B300 faster than the GB300?

At FP8 precision the GB300 reaches 5,000 TFLOPS dense against 4,500 TFLOPS for the B300. The performance table on this page lists every published precision for both GPUs.

Which has more memory, the B300 or the GB300?

The B300 carries 288 GB of VRAM versus 252 GB for the GB300. Memory bandwidth is 8.0 TB/s for the B300 and 7.1 TB/s for the GB300.

Can I rent the B300 or the GB300 in the cloud?

Yes. Live cloud listings tracked by Flopper start at $3.75 per GPU hour. The rental pricing section on this page lists current providers and rates for both GPUs.

Where to Rent

NVIDIA B300 SXM 288GB

ProviderInstancePrice/GPU-hrChecked
DataCrunch
1B300.30V (spot) $3.75 3h ago View →
Nebius
$4.30 3h ago View →
CoreWeave
nvidia-b300 (spot) $4.48 3h ago View →
RunPod
B300 (Community Cloud) $6.94 3h ago View →
Hyperstack
$7.40 3h ago View →
DataCrunch
1B300.30V (on-demand) $7.50 3h ago View →
Nebius
$7.85 3h ago View →
RunPod
B300 (Secure Cloud) $7.89 3h ago View →
Scaleway
B300-SXM-8-288G $8.76 3h ago View →
DigitalOcean
gpu-b300x1-288gb $11.19 3h ago View →
DigitalOcean
gpu-b300x8-2304gb $11.19 3h ago View →
All B300 listings and price history →

Get Comparison Updates

New GPUs added weekly. Be the first to see how they compare.