AI GPU Comparison Chart (2026): Specs & Prices

Specifications and monthly prices of the GPUs OverGPU rents
GPU	Memory	Bandwidth	FP8 dense	FP4 dense	Interconnect	Power	From
B300NVIDIA · Blackwell Ultra	288 GBHBM3e	8 TB/s	4,500 TFLOPS	13,500 TFLOPS	NVLink 51.8 TB/s	≈1,100 W	$5,190/mo · 1 GPU
B200NVIDIA · Blackwell	180 GBHBM3e	8 TB/s	4,500 TFLOPS	9,000 TFLOPS	NVLink 51.8 TB/s	≈1,000 W	$4,590/mo · 1 GPU
H200NVIDIA · Hopper	141 GBHBM3e	4.8 TB/s	1,979 TFLOPS	—	NVLink 4900 GB/s	700 W	$3,090/mo · 1 GPU
H100NVIDIA · Hopper	80 GBHBM3	3.35 TB/s	1,979 TFLOPS	—	NVLink 4900 GB/s	700 W	$2,190/mo · 1 GPU
MI355XAMD · CDNA 4	288 GBHBM3e	8 TB/s	5,000 TFLOPS	10,100 TFLOPSMXFP4	Infinity Fabric	1,400 W	$19,900/mo · 8 GPUs, ≈ $2,488/GPU
RTX PRO 6000NVIDIA · Blackwell	96 GBGDDR7 ECC	1.6 TB/s	Supported	4,000 AI TOPS	PCIe Gen5 x16	600 W	$1,390/mo · 1 GPU
L40SNVIDIA · Ada Lovelace	48 GBGDDR6 ECC	864 GB/s	733 TFLOPS	—	PCIe Gen4 x16	350 W	$890/mo · 1 GPU
RTX 5090NVIDIA · Blackwell	32 GBGDDR7	1.79 TB/s	Supported	3,352 AI TOPS	PCIe Gen5 x16	575 W	$449/mo · 1 GPU
GB300 NVL72NVIDIA · Blackwell Ultra	288 GBHBM3e	8 TB/s	5,000 TFLOPS	15,000 TFLOPS	NVLink 51.8 TB/s	Rack	$7,990/GPU·mo
Vera Rubin NVL72NVIDIA · Rubin	288 GBHBM4	19.2 TB/s	17,500 TFLOPS	35,000 TFLOPSNVFP4	NVLink 6	Rack	On requestReservations open

Swipe the table to see every column

Vendor-published figures per GPU; compute is dense math. Rack-scale systems are shown per GPU. Prices in USD per month, excluding VAT. Download the chart as a CSV file.

Head to head

Detailed comparisons

Data center GPUs

Comparison A100 vs H100: what changed, and why we rent the H100 With the same 80 GB, the H100 adds FP8, 64% more bandwidth, about 3.2 times the vendor-listed dense BF16 compute and faster NVLink, so it is the GPU to rent in 2026, while A100s make sense mainly where you already have them. Read the comparison
Comparison B200 vs B300: is Blackwell Ultra worth the upgrade? For FP8 and BF16 work that fits in 180 GB, the B200 does the same job for less and is in stock; the B300 earns its 13% premium with 288 GB, faster FP4 and 800G networking, if you can plan around limited stock. Read the comparison
Comparison H100 vs B200: is the H100 still worth renting in 2026? The B200 is the better value per unit of work when you can keep it busy, but at $2,190 a month the H100 remains the right buy for models that fit in 80 GB and workloads that would leave a B200 half idle. Read the comparison
Comparison H100 vs H200: same Hopper compute, very different memory If your model and its KV cache fit in 80 GB, the H100 does the same compute for $900 less a month; once they do not, the H200 puts 141 GB on one GPU and becomes the better buy. Read the comparison
Comparison H200 vs B200: Hopper or Blackwell for your next GPU server? Pick the B200 when you can use its compute, bandwidth and 180 GB; pick the H200, which costs $1,500 less per GPU and less per GB, when your model fits in 141 GB and Hopper keeps up with your traffic. Read the comparison
Comparison L40S vs H100: which GPU fits your workload and budget? For small and medium models, media and rendering, the L40S does the job at $890 a month; for training, larger models and bandwidth-hungry LLM serving, the H100 is worth its $2,190. Read the comparison
Comparison MI355X vs B200: memory, software and cost compared If memory is your constraint and your stack runs on ROCm, the MI355X server gives you 2,304 GB of HBM3e for $19,900 a month; if you depend on CUDA-only software or need one GPU, choose the B200. Read the comparison
Comparison MI355X vs H200: double the memory or the proven stack? If your software runs on ROCm, the MI355X server gives you twice the memory and newer precision formats for $4,000 a month less than an HGX H200; for CUDA-only stacks or a single GPU, choose the H200. Read the comparison
Comparison RTX PRO 6000 vs H100: more memory or more bandwidth? For single-GPU inference of large quantized models, the RTX PRO 6000 gives you more memory for less money; for multi-GPU training and bandwidth-bound serving, the H100 is the right tool. Read the comparison

Workstation and consumer GPUs

Pricing models

Monthly vs hourly GPU pricing: how to find your break-even

Divide the monthly price by the hourly rate you would pay: if your GPU will work more hours than that, monthly billing wins; if fewer, pay by the hour. Run your numbers

Comparing providers instead?

How to read this AI GPU comparison chart

Memory Capacity per GPU decides which models, contexts and batch sizes fit at all.
Bandwidth Memory bandwidth sets how fast a GPU generates tokens once a model fits, because decoding reads the weights for every token.
FP8 and FP4 Dense tensor throughput, as the vendor publishes it, drives training and prompt processing. FP4 is native on Blackwell GPUs and on the MI355X.
Interconnect NVLink or Infinity Fabric lets the GPUs of one server work as one memory pool for tensor parallelism.
Price Our flat monthly price per server, with the per-GPU-hour equivalent at 730 hours a month in each comparison.

Which GPU for which workload

Workload	Good fit	Why
Image generation and models up to about 14B	RTX 5090, L40S	32 to 48 GB at the lowest monthly prices
Serving 30B to 70B models	RTX PRO 6000, H200	96 to 141 GB on one GPU
Fine-tuning and training up to 70B	HGX H100, HGX H200	Eight GPUs on NVLink 4 and a mature software stack
The largest models, FP4 and big batches	B200, B300, MI355X	180 to 288 GB per GPU and native FP4
More than eight GPUs in one NVLink domain	GB300 NVL72	72 GPUs in one rack

To check a specific model, the LLM GPU requirements pages list every configuration that runs it, and the VRAM calculator sizes any other.

How this chart is made

Specifications are the figures NVIDIA and AMD publish for each GPU: memory capacity and bandwidth, dense FP8 and FP4 throughput without sparsity, interconnect and maximum power. Prices are our current monthly prices, taken from the same catalog as the order page. Each detailed comparison adds what fits in memory, the price per GPU-hour and when to choose each GPU. Terms such as HBM3e, NVLink and FP4 are explained in the GPU glossary.

Check your model first

See the memory your model needs and which of these GPUs hold it, with monthly prices.

Open the VRAM calculator See all monthly prices

FAQ

GPU comparison FAQ

The short version. The help center has the details, and customers with an active server get answers by ticket, 24/7.

Visit the help center

Which GPU is best for LLM inference?

The one whose memory holds your model, its KV cache and some headroom at the lowest price. For 70B models in FP8, an H200 serves 8 concurrent 8k requests on one GPU; smaller models fit an RTX 5090, L40S or RTX PRO 6000.

What is the difference between the H100 and the H200?

Both are Hopper GPUs with the same published compute. The H200 has 141 GB of HBM3e at 4.8 TB/s, against 80 GB of HBM3 at 3.35 TB/s on the H100, so it holds larger models and longer contexts.

Is the B300 better than the B200?

The B300 (Blackwell Ultra) has 288 GB of HBM3e against 180 GB and a higher dense FP4 figure in NVIDIA's published specifications, and HGX B300 servers use 800G networking. If your model fits in 180 GB, the B200 costs less per month.

Should I compare GPUs by price per hour?

Compare by what your work needs first: memory, then bandwidth or compute. Among the GPUs that fit, compare the monthly price at your real usage; the GPU cost calculator converts any hourly quote.

AI GPU comparison chart (2026)