Geekbench AI Score

33496
Single Precision Score
46774
Half Precision Score
42837
Quantized Score
Geekbench AI 1.7.0 for macOS AArch64

Result Information

User SeaRefractor
Upload Date October 21 2026 07:33 PM
Views 15

System Information

AI Information
Framework Core ML
Backend GPU
Device Apple M5 Ultra
System Information
Operating System macOS 27.0.1 (Build 26A434)
Model Mac17,15
Model IDMac17,15
Motherboard Mac17,15
CPU Information
Name Apple M5 Ultra
Topology 1 Processor, 30 Cores
Identifier Apple M5 Ultra
Base Frequency 4.60 GHz
Cluster 110 Cores
Cluster 220 Cores
Memory Information
Size 96.00 GB

AI Performance

Workload Accuracy Score
Image Classification (SP)
100% 11420
2.12 KIPS
 
Image Classification (HP)
100% 12422
2.31 KIPS
 
Image Classification (Q)
99% 12435
2.32 KIPS
 
Image Segmentation (SP)
100% 37950
615.2 IPS
 
Image Segmentation (HP)
100% 52543
851.8 IPS
 
Image Segmentation (Q)
99% 49160
796.9 IPS
 
Pose Estimation (SP)
100% 161816
188.8 IPS
 
Pose Estimation (HP)
100% 380669
444.2 IPS
 
Pose Estimation (Q)
96% 336009
393.7 IPS
 
Object Detection (SP)
100% 11108
881.1 IPS
 
Object Detection (HP)
100% 12656
1.00 KIPS
 
Object Detection (Q)
92% 13773
1.10 KIPS
 
Face Detection (SP)
100% 61305
728.4 IPS
 
Face Detection (HP)
100% 75521
897.4 IPS
 
Face Detection (Q)
100% 65424
777.4 IPS
 
Depth Estimation (SP)
100% 55733
429.4 IPS
 
Depth Estimation (HP)
99% 70412
542.5 IPS
 
Depth Estimation (Q)
85% 73352
573.2 IPS
 
Style Transfer (SP)
100% 384088
493.8 IPS
 
Style Transfer (HP)
100% 767136
986.2 IPS
 
Style Transfer (Q)
96% 660999
853.0 IPS
 
Image Super-Resolution (SP)
100% 36702
1.36 KIPS
 
Image Super-Resolution (HP)
100% 47094
1.74 KIPS
 
Image Super-Resolution (Q)
99% 28799
1.06 KIPS
 
Text Classification (SP)
100% 6405
8.55 KIPS
 
Text Classification (HP)
99% 10701
14.3 KIPS
 
Text Classification (Q)
93% 10712
14.4 KIPS
 
Machine Translation (SP)
100% 7399
127.5 IPS
 
Machine Translation (HP)
100% 7754
133.6 IPS
 
Machine Translation (Q)
98% 7517
129.9 IPS
 

Compare

Set Baseline Geekbench AI now available