NVIDIA B100, AI & Machine Learning Benchmarks & Specs

192GB · AI Score 67.0/100 · anchored estimate vs 51 measured cards

67 AI Score Includes estimates

We have not run NVIDIA B100 on our bench. These figures are anchored estimates, interpolated per workload against the 51 GPUs we did measure (confidence: high (sibling silicon)). On Llama 3.1 8B (Q4_K_M) NVIDIA B100 should deliver about 260.7 tokens/sec. Stepping up to Qwen3 32B it should hold roughly 74.6 tok/s. The full Llama 3.3 70B still runs, at about 42.3 tok/s. For image generation, SDXL should run near 18.45 it/s, and FLUX.1-dev at 4.9 it/s. All 12 workloads fit in 192GB. There is no model in our suite this card has to turn down.

AI & Machine Learning benchmark results

Text Generation tok/s 5

Qwen3 4B302
Llama 3.1 8B260.7
Qwen2.5-Coder 14B143.4
Qwen3 32B74.6
Llama 3.3 70B42.3
WorkloadResultTelemetryData
Qwen3 4B302 tok/sestimatedEst.
Llama 3.1 8B260.7 tok/sestimatedEst.
Qwen2.5-Coder 14B143.4 tok/sestimatedEst.
Qwen3 32B74.6 tok/sestimatedEst.
Llama 3.3 70B42.3 tok/sestimatedEst.

Image Generation images/min 3

Stable Diffusion XL36.9
Z-Image Turbo25.875
FLUX.1 dev10.5
WorkloadResultTelemetryData
Stable Diffusion XL36.9 images/minestimatedEst.
Z-Image Turbo25.88 images/minestimatedEst.
FLUX.1 dev10.5 images/minestimatedEst.

Image Editing images/min 2

WorkloadResultTelemetryData
FLUX.1 Kontext dev4.59 images/minestimatedEst.
Qwen-Image-Edit3.88 images/minestimatedEst.

Video Generation frames/s 2

WorkloadResultTelemetryData
LTX-Video (distilled)21.44 frames/sestimatedEst.
Wan 2.2 5B (720p)1.5 frames/sestimatedEst.
How this estimate is derived. This card hasn’t been through our bench yet, so its numbers are anchored estimates, interpolated from the 51 first-party measured cards (Sibling to measured B200: power-capped (700W vs 1000W), LLM ×0.95 (bandwidth-similar), diffusion ×0.80 (compute-limited)). The VRAM “won’t fit” gates are exact, since they’re pure capacity limits. Confidence: high (sibling silicon). Estimates are replaced with measured data as more silicon goes through the bench. Full methodology →

NVIDIA B100 specifications

ArchitectureBlackwell
VRAM192GB HBM3e
Memory bus8192-bit
Memory bandwidth8 TB/s
TDP700 W
ProcessTSMC 4NP
InterfaceSXM6
Release date2025-01-01
Launch MSRP$32,500

Verdict, nothing in our suite slows it down

NVIDIA B100 scores 67.0/100, #3 of 102. It ran all 12 workloads. Figures are anchored estimates, not measurements, we flag that on every row.

Relative performance: where the NVIDIA B100 lands

100% = this card, AI & Machine Learning headline metric (AI Score). #3 of 21 desktop cards in this vertical.

GPURelative%AI Score
NVIDIA B300
140%93.8
NVIDIA B200
116%78
NVIDIA B100
100%67
NVIDIA H100 NVL
100%67
NVIDIA GH200 Grace Hopper
98%65.6
NVIDIA H200
97%65
NVIDIA H100 80GB HBM3
93%62.6

← All AI & Machine Learning GPU rankings

What this card can build

Whole-job timings, composed from our measured per-model results on this card.

WorkflowTimeEnergyBasis
24-frame storyboard74 sn/aestimate, 0 of 2 stages measured
60-second AI short film1.7 minn/aestimate, 0 of 3 stages measured
6-panel comic page2 minn/aestimate, 0 of 3 stages measured
Character sheet, 12 poses2.7 minn/aestimate, 0 of 2 stages measured
10 short social clips5.9 minn/aestimate, 0 of 3 stages measured
Full codebase review7 minn/aestimate, 0 of 1 stage measured
40-product photo shoot9.8 minn/aestimate, 0 of 2 stages measured
20 long-form articles11 minn/aestimate, 0 of 1 stage measured
100-photo restoration batch21.8 minn/aestimate, 0 of 1 stage measured