H100 vs L4
The H100 and the L4 are both data centre GPUs we supply tested from Hong Kong, and the figures below are the same ones published on each product page. They share the same supply and warranty terms and separate on memory, power, architecture, so the decision is less about "which is better" and more about which of those differences your platform or workload can actually use.
Nothing on this page is a price: used and refurbished hardware is repriced weekly, so both are quoted on request against confirmed stock with the date the quote holds until. If you are replacing a part, send the platform and we will confirm the population before you commit.
Side by side
H100 vs L4 specifications
| Specification | H100 | L4 |
|---|---|---|
| Memory | 80GB HBM3 3.35 TB/s (SXM5) / 80GB HBM2e 2.0 TB/s (PCIe) | 24GB GDDR6 ECC, 300 GB/s |
| Power | 700W (SXM5) / 350W (PCIe), passive | 72W TDP — no external power connector |
| Interface | not stated | PCIe Gen4 x16, single-slot low-profile, passive |
| Architecture | Hopper GH100, 132 SM (SXM5) / 114 SM (PCIe) | Ada Lovelace, 7,424 CUDA cores, 232 Tensor cores |
| Compute | FP8 1,979 TFLOPS / FP16 989 TFLOPS (SXM5, with sparsity) | FP8 242 TFLOPS / FP16 121 TFLOPS / TF32 60 TFLOPS |
| MIG | Up to 7 instances (10GB each) | not stated |
| NVLink | 900 GB/s GPU-to-GPU (SXM5); NVLink bridge (PCIe) | not stated |
| Virtualization | not stated | vGPU support, NVIDIA AI Enterprise ready |
| Model | NVIDIA H100 80GB — SXM5 (900-21010-0000-000) and PCIe variants | NVIDIA L4 (900-2G193-0000-000) |
| Transformer Engine | FP8 training and inference, 4th-gen Tensor cores | not stated |
| Media | not stated | 2x NVENC, 4x NVDEC, 4x JPEG — AV1 encode |
Rows that differ are highlighted. Figures are the same ones published on each product page; where a figure is community-sourced rather than vendor-published it is stated on that page. Confirm every part against your build sheet when ordering.
Where they differ
- ·Memory — H100: 80GB HBM3 3.35 TB/s (SXM5) / 80GB HBM2e 2.0 TB/s (PCIe); L4: 24GB GDDR6 ECC, 300 GB/s.
- ·Power — H100: 700W (SXM5) / 350W (PCIe), passive; L4: 72W TDP — no external power connector.
- ·Architecture — H100: Hopper GH100, 132 SM (SXM5) / 114 SM (PCIe); L4: Ada Lovelace, 7,424 CUDA cores, 232 Tensor cores.
- ·Compute — H100: FP8 1,979 TFLOPS / FP16 989 TFLOPS (SXM5, with sparsity); L4: FP8 242 TFLOPS / FP16 121 TFLOPS / TF32 60 TFLOPS.
Decision
Which one to buy
Data centre GPUs are chosen by memory footprint per job and by the deployment you already have. More memory per card means bigger models, larger batches and fewer cards for the same workload; less memory per card is often the better buy when you scale out across many nodes. Memory bandwidth and compute format matter for the workload — FP8 and BF16 throughput for inference, FP64 for simulation — and the power envelope decides how many cards fit per chassis. Check the form factor and cooling as well: a passive server card and an actively cooled workstation card are not interchangeable.
Choose the H100 if
- ·Memory: 80GB HBM3 3.35 TB/s (SXM5) / 80GB HBM2e 2.0 TB/s (PCIe)
- ·Power: 700W (SXM5) / 350W (PCIe), passive
- ·Architecture: Hopper GH100, 132 SM (SXM5) / 114 SM (PCIe)
Choose the L4 if
- ·Memory: 24GB GDDR6 ECC, 300 GB/s
- ·Power: 72W TDP — no external power connector
- ·Architecture: Ada Lovelace, 7,424 CUDA cores, 232 Tensor cores
Frequently asked questions
How much do the H100 and the L4 cost?
We do not publish prices for either: secondary-market hardware is repriced weekly. Send the quantity, the platform and the condition you want and we quote both against confirmed stock, with the date the figure holds until.
What is the main difference between the H100 and the L4?
Memory (H100: 80GB HBM3 3.35 TB/s (SXM5) / 80GB HBM2e 2.0 TB/s (PCIe); L4: 24GB GDDR6 ECC, 300 GB/s); Power (H100: 700W (SXM5) / 350W (PCIe), passive; L4: 72W TDP — no external power connector); Architecture (H100: Hopper GH100, 132 SM (SXM5) / 114 SM (PCIe); L4: Ada Lovelace, 7,424 CUDA cores, 232 Tensor cores).
Can I use the H100 and the L4 in the same system?
They can share a server if the power and slot budget allows it, but they target different jobs: H100 and L4 differ mainly on Memory and Power. Mixed GPU models in one host is a driver/CUDA compatibility decision, not a hardware one.
Which one should I buy?
Buy the one whose strong points match the job: see the "Choose" section above. If you are replacing an existing part, send the platform or the old part number and we will confirm compatibility before you order.
What condition are they in, and do you ship worldwide?
Both are supplied tested with a 1-year written warranty. Condition is stated line by line before you order — new, decommissioned pull or OEM tray — and we ship worldwide from the Hong Kong warehouse with full export documentation, typically 3–7 days by air.
Get a quote for either — or both
Tell us the quantity, the platform and the condition you need. We confirm stock and put a firm figure in writing with the date it holds until.
Request a Quote