NVIDIA B200 GPU rental

Blackwell architecture with 180 GB of HBM3e per GPU. For large language model inference, fine-tuning and distributed training on compatible HGX systems.

Check the console or contact sales for prices and available instance configurations.

GPU memory
180 GB HBM3e
Memory bandwidth
7.7 TB/s
Architecture
Blackwell
Core capabilities

For large-model training and inference

B200 combines high-capacity HBM3e, second-generation Transformer Engine and fifth-generation NVLink. Size the system for model state, numerical precision and GPU communication.

180 GB HBM3e memory

Keep model weights, KV cache and working data in GPU memory. Context length, concurrency and training state determine the remaining capacity required.

FP8 and FP4 Tensor Cores

Second-generation Transformer Engine supports low-precision training and inference. FP4 and FP8 use requires compatible models, quantization, frameworks and kernels.

Fifth-generation NVLink

HGX B200 connects eight GPUs through NVLink and NVSwitch, with 1.8 TB/s of aggregate bidirectional bandwidth per GPU. Cross-node networking is a separate configuration.

Two dies, one CUDA GPU

Blackwell connects two dies through a 10 TB/s chip-to-chip interface and presents them as one CUDA GPU. This internal link is separate from NVLink between GPUs.

Workloads

What can you run on B200?

From model serving to distributed training, select resources for the workload's memory, compute precision and GPU communication needs.

Evaluate quantized LLM or MoE serving against the 180 GB HBM3e budget for weights, KV cache and runtime buffers. Select FP4 or FP8 only after checking kernel support and model accuracy.

For fine-tuning and other post-training jobs, budget activations, gradients and optimizer state alongside weights. Choose GPU count and parallelism from the complete training state, with storage planned for checkpoints.

When a model spans GPUs, evaluate the eight-GPU HGX topology and 1.8 TB/s bidirectional NVLink per GPU. Extending the job across nodes also requires a separately confirmed cluster network and communication strategy.

For text, image or video models, include encoder activations, input resolution and sequence length in the 180 GB memory budget. Verify operator and precision support before expanding image size, frame count or batch size.

Technical specifications

HGX B200 per-GPU reference specifications

The figures below are per GPU, not aggregate specifications for a complete multi-GPU system. Theoretical peaks differ from measured model performance and do not describe a complete instance's CPU, system RAM, storage or cross-node networking.

HGX B200 per-GPU reference specifications
SpecificationValue per GPU
ArchitectureBlackwellBlackwell GPU architecture.
GPU memory180 GB HBM3eStores model weights, KV cache and runtime state.
Memory bandwidth7.7 TB/sData transfer bandwidth between the GPU and local HBM3e memory.
FP8 Tensor compute · Dense4.5 PFLOPS · DenseTheoretical peak for dense Tensor Core operations.
FP8 Tensor compute · Sparse9 PFLOPS · SparseTheoretical peak under the applicable structured-sparsity conditions.
GPU interconnect1.8 TB/s · NVLink 5Aggregate bidirectional GPU-to-GPU bandwidth within compatible HGX systems.

Deployment configuration

Before renting, confirm CPU, system memory, storage, networking, deployment region, pricing, and delivery arrangements.

GPU and host resources
Check GPU count, available GPU memory, CPU cores and system RAM per instance against your model, data preparation and concurrency needs.
Storage and interconnect
Confirm storage type, capacity and persistence. For multiple GPUs or nodes, also check the GPU interconnect, network bandwidth and topology.
Region and rental period
Match the deployment region, start date and rental period to your data location and usage plans. Confirm availability for the chosen configuration.
Pricing and delivery
Review the billing unit, included resources, delivery schedule and access method. Confirm whether storage, network traffic and other charges are included in the quote.
  • SOC 2 Type II audited

  • ISO/IEC 27001 certified

  • GDPR compliant

Find the GPU for your workload

Compare rental options, explore GPU specifications, then discuss configuration and delivery with our team.

Rent a GPU nowNVIDIAB300Explore B300 specificationsNVIDIAH200Explore H200 specificationsNVIDIAH100Explore H100 specificationsNVIDIAA100Explore A100 specificationsNVIDIARTX 5090Explore RTX 5090 specifications
B200

What a GPU-hour costs right now.

Check the console or contact sales for prices and available instance configurations.

Market referenceUSD per GPU-hour
$7.49
30d

Check the console or contact sales for prices and available instance configurations.

Selection and rental

Frequently asked questions

Distinguish GPU specifications, system platforms and the actual delivered configuration.

Are HGX B200 and GB200 NVL72 the same platform?

No. HGX B200 is an eight-GPU Blackwell platform for x86 servers. GB200 NVL72 is a rack-scale system with 72 Blackwell GPUs and 36 Grace CPUs. Their CPU architecture, memory specifications and delivery formats differ.

Why does this page show 7.7 TB/s instead of 8 TB/s?

The HGX B200 column in the NVIDIA datasheet used here lists 7.7 TB/s per GPU; another NVIDIA reference architecture lists up to 8 TB/s. This page keeps the datasheet value. Confirm the exact hardware specification when selecting an instance.

Will FP8 or FP4 peak compute equal my model's throughput?

No. The FP8 Dense and Sparse figures are theoretical Tensor Core peaks; Sparse requires the applicable sparsity conditions. FP4 and FP8 also need compatible model formats and kernels. Measure accuracy, latency and throughput with your model and runtime.

How do I request a B200 configuration and quote?

Provide B200 quantity, preferred region, start date, duration and single-node or multi-node requirements on the GPU requirements page. Confirm the specific configuration, capacity, access method and quote before ordering.

Rent NVIDIA B200 GPUs

View configurations, or submit your GPU quantity, region and rental term for a quote.