このページは英語版のみご利用いただけます。 英語版が正式な内容となります。
GPU catalogue
The supported GPU models, their reference prices in MZG per GPU-hour, how live prices relate to them, and how to choose a GPU for your workload.
MZGrid supports thirteen NVIDIA GPU models. Each is identified by a stable sku used in the API, the SDK and provider pricing.
| SKU | GPU | VRAM | アーキテクチャ | FP16 TFLOPS | 参考価格(MZG/h) |
|---|---|---|---|---|---|
| b200 | NVIDIA B200 | 192 GB | Blackwell | 2,250 | 1,500 |
| h200 | NVIDIA H200 SXM | 141 GB | Hopper | 989 | 1,000 |
| h100-sxm | NVIDIA H100 SXM | 80 GB | Hopper | 989 | 750 |
| h100-pcie | NVIDIA H100 PCIe | 80 GB | Hopper | 756 | 640 |
| a100-80 | NVIDIA A100 SXM 80GB | 80 GB | Ampere | 312 | 460 |
| a100-40 | NVIDIA A100 PCIe 40GB | 40 GB | Ampere | 312 | 320 |
| l40s | NVIDIA L40S | 48 GB | Ada | 362 | 290 |
| rtx6000-ada | NVIDIA RTX 6000 Ada | 48 GB | Ada | 364 | 240 |
| rtx5090 | GeForce RTX 5090 | 32 GB | Blackwell | 210 | 220 |
| a6000 | NVIDIA RTX A6000 | 48 GB | Ampere | 155 | 160 |
| rtx4090 | GeForce RTX 4090 | 24 GB | Ada | 165 | 150 |
| rtx4080s | GeForce RTX 4080 SUPER | 16 GB | Ada | 104 | 90 |
| rtx3090 | GeForce RTX 3090 | 24 GB | Ampere | 71 | 75 |
Reference prices vs. live prices
The reference price is a catalogue anchor, not what you pay. Every provider sets its own price per SKU, and the marketplace shows live offers, which typically sit between 0.85× and 1.20× the reference:
| SKU | Reference MZG/h | Typical live range MZG/h |
|---|---|---|
h100-sxm | 750 | 637.50 – 900 |
a100-80 | 460 | 391 – 552 |
l40s | 290 | 246.50 – 348 |
rtx4090 | 150 | 127.50 – 180 |
rtx3090 | 75 | 63.75 – 90 |
Offers at the lower end usually come from newer providers with less reputation or from community-tier machines; datacenter-tier providers tend to price toward the upper end. Current lowest and median prices per SKU are returned by GET /v1/catalog/gpus.
The FP16 reference column
The FP16 dense TFLOPS figure is the benchmark reference used for verification. A provider's measured throughput must fall within 70–115% of this value for the machine to be listed as verified. It is also a useful rough guide to relative compute speed between SKUs.
Choosing a GPU
VRAM first
VRAM decides what fits. As a rule of thumb, model weights need about 2 GB per billion parameters in BF16/FP16, 1 GB in FP8/INT8 and 0.5–0.6 GB with 4-bit quantisation (AWQ/GPTQ). Add headroom for the KV cache, which grows with context length and the number of concurrent requests — plan for at least 20–30% on top of the weights for serving.
8B models
An 8B model in BF16 needs about 16 GB for weights. A 24 GB card (rtx4090, rtx3090) runs it comfortably with a moderate context; rtx5090 (32 GB) or l40s (48 GB) give room for long contexts and many concurrent requests. The RTX 4090 usually offers the best throughput per MZG; the RTX 3090 is the cheapest option.
32B models
A 32B model in 4-bit AWQ needs roughly 18–20 GB, so it fits on 24 GB cards with a short context (qwen2.5-32b template). For production traffic or longer contexts, a 48 GB card (l40s, rtx6000-ada, a6000) is the sweet spot. In BF16 a 32B model needs about 64 GB of weights — use an a100-80 or h100.
70B models
A 70B model in FP8 needs about 70 GB, so use an 80 GB or larger Hopper/Blackwell GPU: h100-sxm, h100-pcie, h200 (141 GB, much more KV-cache room) or b200 (192 GB). This is what the llama-3.3-70b template targets. Hopper and Blackwell run FP8 natively, which is why they are preferred over a100-80 for this template.
Image generation
SDXL and FLUX.1-schnell run well on 16–24 GB cards. The rtx4090 is the usual choice for interactive ComfyUI work; rtx4080s is a cheaper option for SDXL at standard resolutions; l40s or rtx5090 are better for large batches or high resolutions.
Speech and notebooks
Whisper large-v3 needs about 10 GB — an rtx4080s or rtx3090 is plenty. JupyterLab runs on anything; choose the GPU that matches the experiments you plan to run.
Summary
| Workload | Minimum | Good value | High throughput |
|---|---|---|---|
| 7–8B LLM (BF16) | 24 GB | rtx4090, rtx3090 | l40s, h100-pcie |
| 32B LLM (AWQ) | 24 GB | rtx4090, rtx5090 | l40s, rtx6000-ada |
| 70B LLM (FP8) | 80 GB | h100-pcie | h100-sxm, h200, b200 |
| SDXL / FLUX.1-schnell | 16 GB | rtx4080s, rtx4090 | rtx5090, l40s |
| Whisper large-v3 | 10 GB | rtx4080s, rtx3090 | rtx4090 |
| Fine-tuning / research | depends | a100-80 | h100-sxm, h200 |