Hot starts, batch inference, and what's next for Runpod Serverless. Webinar June 25.

Cloud GPUs

Rent NVIDIA B200 GPUs from $5.89/hr

Next-generation data center GPU based on Blackwell architecture that features 192GB of HBM3e memory with 8TB/s bandwidth, delivering up to 20 petaFLOPS of FP4 AI compute performance.

B200

Powering the next generation of AI & high-performance computing.

Engineered for large-scale AI training, deep learning, and high-performance workloads, delivering unprecedented compute power and efficiency.

Built on NVIDIA Blackwell Architecture

Revolutionary dual-die design with 208 billion transistors delivering breakthrough performance for trillion-parameter AI models.

Second-Generation Transformer Engine

Custom Tensor Core technology accelerates LLM inference and training with new FP4 precision for up to 2.5X performance gains.

192GB HBM3e at 8TB/s

Massive memory capacity with ultra-high bandwidth enables training and inference on the largest AI models without memory constraints.

Fifth-Generation NVLink

1.8TB/s GPU-to-GPU interconnect allows seamless scaling across multiple GPUs for distributed AI workloads.

Performance

Key specs at a glance.

Performance benchmarks that push AI, ML, and HPC workloads further.

Memory Bandwidth

8

TB/s

FP16 Tensor Performance

2.2

PFLOPS

NVLink Bandwidth

1.8

TB/s

Use Cases

Popular use cases.

Designed for demanding workloads
—learn if this GPU fits your needs.

Inference workload illustration

Inference

Serve inference for image, text, and audio generation at any scale.

Inference
Fine-tuning workload illustration

Fine-tuning

Train custom models on
your specific datasets.

Fine-tuning
AI agents workload illustration

Agents

Build intelligent agent-based systems and workflows.

Agents
Compute-heavy workload illustration

Compute-heavy tasks

Run compute-heavy workloads like rendering and simulations.

Compute-heavy tasks

Technical Specs

Ready for your most
demanding workloads.

Essential technical specifications to help you choose the right GPU for your workload.

Specification
Details
Great for...
Memory Bandwidth
8 TB/s
Delivering large datasets to GPU memory without stalls for high-resolution simulations and LLM inference.
FP16 Tensor Performance
2.2 PFLOPS
Accelerating mixed-precision transformer training and inference, reducing fine-tuning times and boosting token throughput.
NVLink Bandwidth
1.8 TB/s
Enabling fast GPU-to-GPU communication in multi-card setups, avoiding PCIe bottlenecks for large model scaling.
AI Apps
Research
LLM Inference
Model Training
"The Runpod team has clearly prioritized the developer experience to create an elegant solution that enables individuals to rapidly develop custom AI apps or integrations while also paving the way for organizations to truly deliver on the promise of AI."

Amjad Masad

"Runpod is the only place I can deploy high-end GPU models instantly—no sales calls, no rate limits, no nonsense."

Daniel Chang

“The main value proposition for us was the flexibility Runpod offered. We were able to scale up effortlessly to meet the demand at launch.”

Josh Payne

“Runpod helped us scale the part of our platform that drives creation. That’s what fuels the rest—image generation, sharing, remixing. It starts with training.”

Matty Shimura

Comparison

Powerful GPUs. Globally available.
Reliability you can trust.

30+ GPUs, 31 regions, instant scale. Fine-tune or go full Skynet—we’ve got you.

Community Cloud
$5.89/hr
Secure Cloud
$5.89/hr
Unique GPU Models
Community Cloud
25
Secure Cloud
19
Global Regions
Community Cloud
17
Secure Cloud
14
Network Storage
Community Cloud
Secure Cloud
Enterprise-Grade Reliability
Community Cloud
Secure Cloud
Savings Plans
Community Cloud
Secure Cloud
24/7 Support
Community Cloud
Secure Cloud
Delightful Dev Experience
Community Cloud
Secure Cloud

10,100,100,100

Requests since launch & 400k developers worldwide

Build what’s next.

Build, train, and scale AI workloads on Runpod with cloud GPUs, Serverless, and Clusters.