Specifications Compared
| Spec | GB300 SXM6 | RTX 4090 |
|---|---|---|
| TDP | 1400W | 450W |
| VRAM | 288 GB | 24 GB |
| CUDA Cores | Not published | 16,384 |
| FP4 (dense) | 13,500 TFLOPS | Not published |
| FP8 (dense) | 4,500 TFLOPS | 330.3 TFLOPS |
| Memory Type | HBM3e | GDDR6X |
| Architecture | Blackwell Ultra | Ada Lovelace |
| FP16 (dense) | 2,250 TFLOPS | 165.2 TFLOPS |
| Form Factors | SXM6 | PCIe |
| INT8 (dense) | 187.5 TOPS | 660.6 TOPS |
| Interconnect | NVLink, PCIe 6.0, InfiniBand | PCIe 4.0 |
| Tensor Cores | Not published | 512 |
| FP32 Performance | 75 TFLOPS | 82.6 TFLOPS |
| FP64 Performance | 1.25 TFLOPS | 1.3 TFLOPS |
| Memory Bandwidth | 8,000 GB/s | 1,008 GB/s |
| FP4 (with sparsity) | 18,000 TFLOPS | Not published |
| FP8 (with sparsity) | 9,000 TFLOPS | 660.6 TFLOPS |
| FP16 (with sparsity) | 4,500 TFLOPS | 330.4 TFLOPS |
| INT8 (with sparsity) | 375 TOPS | 1,321.2 TOPS |
Performance Analysis
The GB300 SXM6 variant shares a dense FP16 rating of 2250 TFLOPS and an FP32 rating of 75 TFLOPS with its family while the RTX 4090 variant shares a dense FP16 rating of 165.2 TFLOPS and an FP32 rating of 82.6 TFLOPS with its family. This produces a dense FP16 advantage of roughly 13.6 times for the GB300 SXM6 that supports larger model training workloads and higher inference throughput when sparsity is not applied. Memory bandwidth shared at 8000 GB/s for the GB300 SXM6 family versus 1008 GB/s for the RTX 4090 family allows substantially larger batch sizes during training or inference on the former without swapping to system memory.
Current On-Demand Offers
Cheapest secure on-demand offer per provider that is reported in stock right now, per GPU per hour; multi-GPU instances show the whole-instance price alongside. Deploy opens the DeployGPU console for providers on the platform, otherwise the provider's own site. Stock is rechecked every minute.
GB300 SXM6
GB300 SXM6 is not offered on-demand by any provider we track right now. See the GB300 SXM6 rental page for last-seen listed prices and a price alert.
Notify me when GB300 SXM6 drops below a price
One email when the cheapest in-stock on-demand price per GPU-hour falls below your threshold.
QuantaCloud
Comparing B-series options? Get one quote for all of them.
Skip the per-provider sales calls. Reserved and cluster B-series configurations from 16 to 1024+ GPUs with InfiniBand fabric, 3 to 12 month terms. One quote at partner rates, 24h turnaround.
When to Choose the GB300 SXM6
The GB300 SXM6 variant suits large scale LLM training runs that require the full 288 GB HBM3e capacity and 8000 GB/s shared family bandwidth to hold entire models in memory. Its NVLink interconnect further supports multi GPU scaling for distributed workloads that exceed single device limits.
When to Choose the RTX 4090
The RTX 4090 variant fits smaller inference tasks or fine tuning jobs that operate within the 24 GB GDDR6X capacity and 450W shared family TDP. Its PCIe 4.0 interconnect matches standard workstation or server slots without requiring specialized SXM6 infrastructure.
Use Cases
The GB300 SXM6 variant holds models in its 288 GB HBM3e memory while the RTX 4090 variant cannot accommodate the same scale with only 24 GB GDDR6X.
Dense FP16 throughput of 2250 TFLOPS on the GB300 SXM6 variant supports higher token rates than the 165.2 TFLOPS on the RTX 4090 variant.
Both variants handle moderate parameter counts yet the GB300 SXM6 variant processes larger batches thanks to 8000 GB/s shared bandwidth.
The RTX 4090 variant delivers sufficient 82.6 FP32 TFLOPS and 24 GB capacity for typical image generation workloads at lower 450W TDP.
FP32 performance combined with 8000 GB/s bandwidth on the GB300 SXM6 variant accelerates simulations that exceed 24 GB limits.
Frequently Asked Questions
What memory capacities distinguish the GB300 SXM6 variant from the RTX 4090 variant?▾
The GB300 SXM6 variant contains 288 GB HBM3e while the RTX 4090 variant contains 24 GB GDDR6X. These capacities remain fixed for the named variants across their respective families.
How do the architectures differ between the GB300 SXM6 and RTX 4090 variants?▾
The GB300 SXM6 variant uses Blackwell Ultra architecture released in 2025 while the RTX 4090 variant uses Ada Lovelace architecture released in 2022. All tensor and bandwidth figures derive from these respective architectures.
Which variant supports higher dense FP16 throughput?▾
The GB300 SXM6 variant delivers 2250 dense FP16 TFLOPS as a shared family figure while the RTX 4090 variant delivers 165.2 dense FP16 TFLOPS as a shared family figure.
What interconnect options appear on each variant?▾
The GB300 SXM6 variant includes NVLink alongside PCIe 6.0 and InfiniBand while the RTX 4090 variant uses only PCIe 4.0. These interconnects stay consistent within each family.
How does TDP compare between the two variants?▾
The GB300 SXM6 variant shares a 1400W TDP with its family while the RTX 4090 variant shares a 450W TDP with its family. No variant specific TDP values exist beyond these family figures.
Which is cheaper to rent, the GB300 SXM6 or the RTX 4090?▾
Cloud rental prices for both the GB300 SXM6 and RTX 4090 vary by provider, configuration, and availability. This page shows live pricing from 25+ providers updated every 60 seconds. Scroll to the Live Cloud Pricing section to compare current rates.
How much VRAM does the GB300 SXM6 have compared to the RTX 4090?▾
The GB300 SXM6 has 288 GB of HBM3e memory. The RTX 4090 has 24 GB of GDDR6X memory.
Can I find GB300 SXM6 and RTX 4090 GPUs available to rent right now?▾
Yes. This page shows real-time availability across 25+ cloud GPU providers. The Live Cloud Pricing section displays only in-stock offers with current pricing.
What is the main difference between the GB300 SXM6 and the RTX 4090?▾
The GB300 SXM6 uses the Blackwell Ultra architecture (2025) while the RTX 4090 uses Ada Lovelace (2022). The GB300 SXM6 delivers 13.6x the dense FP16 throughput (2,250 vs 165.2 TFLOPS, both without sparsity) and 7.9x the memory bandwidth of the RTX 4090.
Rent these GPUs
Each GPU page lists every current on-demand offer by provider, per GPU-hour, updated every minute.
Related comparisons
How this page is made
- Specifications come from the NVIDIA, AMD and Intel datasheets for the GB300 SXM6 and the RTX 4090. Dense and sparse throughput are listed separately.
- Prices and availability are live: every offer shown is an in-stock, on-demand listing from a provider we track, rechecked every minute.
- The written comparison (overview, performance notes, when to choose each card, verdict, use cases and FAQ) was drafted with an AI model from the specification table above and passed an automated check that rejects any figure not in that table. It was last generated on . It contains no prices; those are always read live.
- Read how we collect the data, or report an error on this page.
Next steps
- Rent GB300Every current offer by provider, daily price history and a price alert.
- Rent RTX 4090Every current offer by provider, daily price history and a price alert.
- GPU price indexHow on-demand prices for the major GPUs have moved, updated daily.
- All GPUsEvery GPU we track, with current lows and provider counts.