Specifications Compared
| Spec | B300 | L40S |
|---|---|---|
| TDP | 1400W | 350W |
| VRAM | 262-288 GB | 48 GB |
| CUDA Cores | Not published | 18,176 |
| FP4 (dense) | 13,500 TFLOPS | Not published |
| FP8 (dense) | 4,500 TFLOPS | 733 TFLOPS |
| Memory Type | HBM3e | GDDR6 |
| Architecture | Blackwell Ultra | Ada Lovelace |
| FP16 (dense) | 2,250 TFLOPS | 362.05 TFLOPS |
| Form Factors | SXM6 | PCIe |
| INT8 (dense) | 187.5 TOPS | 733 TOPS |
| Interconnect | NVLink, PCIe 6.0, InfiniBand | PCIe 4.0 |
| Tensor Cores | Not published | 568 |
| FP32 Performance | 75 TFLOPS | 91.6 TFLOPS |
| FP64 Performance | 1.25 TFLOPS | 1.4 TFLOPS |
| Memory Bandwidth | 8,000 GB/s | 864 GB/s |
| FP4 (with sparsity) | 18,000 TFLOPS | Not published |
| FP8 (with sparsity) | 9,000 TFLOPS | 1,466 TFLOPS |
| FP16 (with sparsity) | 4,500 TFLOPS | 733 TFLOPS |
| INT8 (with sparsity) | 375 TOPS | 1,466 TOPS |
Performance Analysis
The B300 FP16 dense performance of 2250 TFLOPS exceeds the L40S FP16 dense performance of 362.05 TFLOPS by a factor derived directly from those two numbers. This difference affects training throughput because higher dense FP16 capacity supports larger model computations per cycle. The B300 FP16 with sparsity of 4500 TFLOPS likewise exceeds the L40S FP16 with sparsity of 733 TFLOPS. Memory bandwidth of 8000 GB/s on the B300 versus 864 GB/s on the L40S determines feasible batch sizes because data movement rates limit how many tokens or samples can reside in active computation at once. FP32 figures show the L40S at 91.6 TFLOPS and the B300 at 75 TFLOPS so precision choices must align with the reported dense or sparsity labels.
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.
B300
1 provider in stock, 1 offer. All B300 offers, price history and alerts
L40S
| Provider | Region | GPUs | Per GPU / hr | Instance / hr | Deploy |
|---|---|---|---|---|---|
| Massed Compute | us-central-2 | 4 | $0.97 | $3.88 | Deploy |
| QuantaCloud | us-midwest-1 | 8 | $1.09 | $8.72 | Deploy |
| Ori | lille-4 | 1 | $1.55 | — | Deploy |
| Lyceum | Europe | 1 | $1.99 | — | Deploy |
4 providers in stock, 17 offers (cheapest per provider shown). All L40S offers, price history and alerts
Notify me when B300 drops below a price
One email when the cheapest in-stock on-demand price per GPU-hour falls below your threshold. Cheapest right now: $0.50/GPU-hr.
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 B300
The B300 suits large scale LLM training because its 262-288 GB HBM3e VRAM and 8000 GB/s bandwidth accommodate models that exceed 48 GB capacity. Its FP8 dense rating of 4500 TFLOPS further supports extended inference runs where sparsity can be applied consistently.
When to Choose the L40S
The L40S suits environments where TDP must remain at 350W and form factor is limited to PCIe. Its FP32 rating of 91.6 TFLOPS provides an advantage in workloads that rely on that specific precision without requiring the higher bandwidth of 8000 GB/s.
Use Cases
The B300 supplies 2250 TFLOPS FP16 dense and 262-288 GB VRAM which exceed the L40S equivalents for large model training runs.
The B300 FP8 dense rating of 4500 TFLOPS and 8000 GB/s bandwidth support higher throughput inference compared with the L40S figures.
The B300 4500 TFLOPS FP16 with sparsity enables efficient fine-tuning passes that benefit from its higher memory capacity over the L40S 48 GB limit.
The L40S TDP of 350W and PCIe form factor fit power constrained or single card inference setups where 733 TFLOPS FP16 with sparsity suffices.
The L40S FP32 rating of 91.6 TFLOPS exceeds the B300 value of 75 TFLOPS for precision sensitive scientific workloads.
Frequently Asked Questions
What is the memory difference between B300 and L40S?▾
The B300 provides 262-288 GB HBM3e VRAM while the L40S provides 48 GB GDDR6 VRAM. This difference directly limits the maximum model size each GPU can hold without partitioning.
How does FP16 performance compare on B300 versus L40S?▾
The B300 delivers 2250 TFLOPS FP16 dense and 4500 TFLOPS FP16 with sparsity. The L40S delivers 362.05 TFLOPS FP16 dense and 733 TFLOPS FP16 with sparsity.
What TDP values apply to each GPU?▾
The B300 TDP is listed at 1400W and the L40S TDP is listed at 350W. These values determine cooling and power infrastructure requirements.
Which GPU offers higher memory bandwidth?▾
The B300 lists 8000 GB/s memory bandwidth while the L40S lists 864 GB/s memory bandwidth. Bandwidth affects achievable batch sizes during training or inference.
What interconnect options exist for the B300 and L40S?▾
The B300 supports NVLink, PCIe 6.0, and InfiniBand. The L40S supports PCIe 4.0 only.
Does the B300 support FP4 operations?▾
The B300 lists FP4 dense at 13500 TFLOPS and FP4 with sparsity at 18000 TFLOPS. The L40S specifications do not include FP4 figures.
Which is cheaper to rent, the B300 or the L40S?▾
Cloud rental prices for both the B300 and L40S 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 B300 have compared to the L40S?▾
The B300 has 262 to 288 GB of HBM3e memory. The L40S has 48 GB of GDDR6 memory.
Can I find B300 and L40S 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 B300 and the L40S?▾
The B300 uses the Blackwell Ultra architecture (2025) while the L40S uses Ada Lovelace (2023). The B300 delivers 6.2x the dense FP16 throughput (2,250 vs 362.05 TFLOPS, both without sparsity) and 9.3x the memory bandwidth of the L40S.
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 B300 and the L40S. 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 B300Every current offer by provider, daily price history and a price alert.
- Rent L40SEvery 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.