Model Performance Benchmark

Vision Model Performance

Track how different chips perform on image classification, detection, segmentation, multimodal vision, and related inference workloads.

Benchmark Focus

CV models, multimodal vision, and image-heavy inference pipelines

Comparison Coverage

4 models / 6 profiles

Choose one model and one shared test condition to compare chips side by side

Tracked Models

Published vision models with active benchmark records

Comparable Profiles

Core filters resolve into one shared benchmark profile before table rendering

Chips Covered

Cloud and edge accelerators with published benchmark results

Comparison controls

Pick one model, refine high-frequency conditions in the core area, then use advanced filters only when you need deeper control.

Matching profiles 1

Model

Core Conditions

TaskInput ResolutionBatch SizePrecisionQuantizationFramework

Advanced Conditions

Expose lower-frequency benchmark knobs without crowding the default comparison flow.

SegmentManufacturer

Comparing 5 chip results

Selected model

ResNet-50

INT8 224x224 · Microsoft

Primary metric: throughput (images/s)

Task image-classificationFramework OpenVINORuntime OpenVINO runtimeBatch 256Precision INT8Quant INT8Variant INT8 224x224Subtype image classificationInput 224x224Channels 3

Chip	Segment	Manufacturer	throughput	Throughput	Latency P50	Latency P99	Power	Memory	Source
NVIDIA H200 SXM 989 TFLOPS	Cloud	NVIDIA	18,400 images/s	18,400 images/s	9 ms	12 ms	600 W	28 GB	View Source
Unknown chip —	Edge	—	2,410 images/s	2,410 images/s	26 ms	32 ms	52 W	14 GB	View Source
Unknown chip —	Edge	—	2,280 images/s	2,280 images/s	28 ms	35 ms	22 W	8 GB	View Source
Unknown chip —	Edge	—	1,760 images/s	1,760 images/s	33 ms	42 ms	20 W	6 GB	View Source
Unknown chip —	Edge	—	1,380 images/s	1,380 images/s	41 ms	52 ms	17 W	7 GB	View Source

Loading benchmark controls...

Browse Cloud Chips Browse Edge Chips