B200 192GB
Latest Blackwell data center GPU. FP4 support, massive VRAM, 2nd-gen Transformer Engine.
B200 192GB Overview
The NVIDIA B200 192GB is a Blackwell data-center GPU built for frontier training and large-scale inference. Its 192GB of HBM3e, 3,500 TFLOPS of FP16/BF16 performance, and second-generation Transformer Engine represent a major step beyond Hopper, including lower-precision FP4 capability for compatible models.
With 8,000 GB/s of memory bandwidth, the B200 is designed to keep enormous model weights and KV caches moving at high throughput. The 192GB capacity can place very large quantized models or substantial context windows on one accelerator, while NVLink 5.0 at 1,800 GB/s supports tightly coupled multi-GPU scaling.
It is aimed at next-generation LLM training, ultra-large inference, and scientific workloads that can exploit a Blackwell cluster. The 1,000W power requirement, specialized fabric, and early supply constraints make it disproportionate for ordinary small-model inference.
Memory
Compute Performance
Hardware
Relative Performance
Relative to highest-spec GPU in database
Limitations
Live Cloud PricingOn-demand hourly rates
Compare B200 192GB vs…
Use Case Guidance
LLM Model Size Guidance
Estimates only. Actual capacity depends on context length, KV cache, and framework overhead.
Related Guides
LLM APIs Running on This GPU Class
Providers that serve frontier LLM inference on Blackwell-class hardware.
Related GPUs
Frequently Asked Questions
How much VRAM does the B200 192GB have?
The B200 192GB has 192GB of HBM3e memory with 8000 GB/s bandwidth. This enables running models up to approximately 384B parameters at INT4 precision, 192B at INT8, or 96B at FP16.
What is the FP16 performance of the B200 192GB?
The B200 192GB delivers 3500 TFLOPS of FP16 performance and 3500 TFLOPS BF16, and 7000 TFLOPS FP8. INT8 throughput is 7000 TOPS. For transformer inference, memory bandwidth (8000 GB/s) is often the binding constraint rather than raw TFLOPS.
What is the B200 192GB best used for?
The B200 192GB is best suited for: Next-gen LLM training, Ultra-large model inference, Scientific simulation. Latest Blackwell data center GPU. FP4 support, massive VRAM, 2nd-gen Transformer Engine.
What interconnect does the B200 192GB use?
The B200 192GB uses NVLink 5.0 / PCIe 6.0 with 1800 GB/s NVLink bandwidth for multi-GPU configurations. NVLink enables near-linear tensor-parallel scaling across multiple cards for models that exceed single-card VRAM.
What LLM model sizes can the B200 192GB run?
With 192GB of HBM3e, the B200 192GB can run models up to approximately 96B parameters at FP16 (2 bytes/param), 192B at INT8 (1 byte/param), or 384B at INT4/GGUF (0.5 bytes/param). These are estimates — actual capacity depends on context length, KV cache size, and framework overhead. Longer context windows require more KV cache memory, reducing the effective model size that fits.
How does the B200 192GB compare to the A100 for LLM inference?
The B200 192GB has 3500 TFLOPS FP16 vs the A100 80GB's 312 TFLOPS, and 8000 GB/s memory bandwidth vs the A100's 2,039 GB/s. For memory-bound autoregressive LLM inference, bandwidth is the primary determinant of tokens-per-second. The B200 192GB's higher bandwidth gives it a throughput advantage for large model inference.
What is the power consumption of the B200 192GB?
The B200 192GB has a TDP (Thermal Design Power) of 1000W. This is the maximum sustained power draw under full load. For data center deployments, total rack power consumption is typically 1.2–1.5× the GPU TDP when accounting for CPU, memory, networking, and cooling overhead. At 1000W, the B200 192GB is in the high-power tier — requires specialized data center infrastructure with high-density power delivery.
Ready to rent?
Compare B200 192GB prices across 97+ providers
Live on-demand & spot rates · monthly cost estimates · availability status