Hardware-First AI Evaluation
AI Hardware Performance Rankings
Ranked by NPU TOPS performance - the raw computational capacity that determines on-device AI processing potential.
Understanding NPU Performance
NPU TOPS (Trillions of Operations Per Second) measures the raw computational throughput of a device's neural processing unit. Higher TOPS values indicate stronger on-device AI processing potential for tasks like image generation, voice recognition, and real-time translation.
📊 What TOPS Measures
- Raw computational throughput: Theoretical maximum AI operations per second
- Hardware capacity: Available processing power for AI workloads
- Future potential: Headroom for more demanding AI applications
❌ What TOPS Does NOT Measure
- Software optimization: How well AI features are implemented
- Memory bandwidth: Speed of data moving to/from NPU
- Thermal design: Real-world sustained performance
- Ecosystem integration: Quality of AI features and workflows
- Actual user experience: Real-world performance depends on many factors
Why the #1 device ranks first: The top-ranked device has the highest recorded NPU TOPS score, indicating the greatest raw computational capacity for on-device AI. This provides more headroom for current and future AI applications, though real-world performance also depends on software optimization and thermal design.
🔬 Research Quality Note
Hardware rankings are based on recorded NPU TOPS specifications. For complete AI capability assessment including software support, execution modes, and privacy features, see Overall AI Rankings.
🥇 Best Overall Hardware
Top AI Device for Hardware Performance
iPhone 17 Pro Max
Apple Intelligence
140.0 TOPS
Unmatched NPU performance with 140.0 TOPS. This device provides the strongest hardware foundation for advanced on-device AI workloads, offering the most headroom for future AI applications.
Why This Device Ranks Highly
Outstanding NPU Performance
With 140.0 TOPS, this device has the strongest hardware foundation for on-device AI processing in its category.
Strong Local AI
The Apple Intelligence platform provides on-device AI capabilities with hardware acceleration for AI workloads.
Excellent Hardware Foundation
High TOPS performance provides more headroom for current AI applications and future AI features that may be added through software updates.
| Rank | Device | Platform | NPU TOPS | Research Confidence |
|---|---|---|---|---|
| 1 | iPhone 17 Pro Max | Apple Intelligence | 140.0 | Research required |
| 2 | iPhone 17 Pro | Apple Intelligence | 140.0 | Research required |
| 3 | iPhone Air | Apple Intelligence | 140.0 | Research required |
| 4 | OnePlus 13R | OnePlus AI | 45.0 | Research required |
| 5 | Xiaomi 14T | Xiaomi HyperAI + Google AI | 45.0 | Research required |
| 6 | Xiaomi 15 Ultra | Xiaomi HyperAI | 45.0 | Research required |
| 7 | Pixel 9 | Gemini Nano | 45.0 | Research required |
| 8 | Galaxy Z Fold6 | Galaxy AI | 45.0 | Research required |
| 9 | Galaxy Z Flip6 | Galaxy AI | 45.0 | Research required |
| 10 | Galaxy S24 | Galaxy AI | 45.0 | Research required |
| 11 | Galaxy S25 Ultra | Galaxy AI | 45.0 | Research required |
| 12 | Galaxy S24 Ultra | Galaxy AI | 45.0 | Research required |
| 13 | Galaxy S25 | Galaxy AI | 45.0 | Research required |
| 14 | Galaxy S25+ | Galaxy AI | 45.0 | Research required |
| 15 | OnePlus 13s | OnePlus AI | 45.0 | Research required |
| 16 | Pixel 9 Pro Fold | Gemini Nano | 45.0 | Research required |
| 17 | Xiaomi 15 Pro | Xiaomi HyperAI | 45.0 | Research required |
| 18 | Galaxy Z Fold8 | Galaxy AI | 45.0 | Research required |
| 19 | Galaxy Z Fold7 | Galaxy AI | 45.0 | Research required |
| 20 | Galaxy Z Fold8 Ultra | Galaxy AI | 45.0 | Research required |
| 21 | Galaxy Z Flip7 | Galaxy AI | 45.0 | Research required |
| 22 | Xiaomi 14 | Xiaomi HyperAI | 45.0 | Research required |
| 23 | Xiaomi 15 | Xiaomi HyperAI | 45.0 | Research required |
| 24 | Honor Magic7 | MagicOS AI | 45.0 | Research required |
| 25 | Honor Magic7 Pro | MagicOS AI | 45.0 | Research required |
| 26 | OnePlus 15 | OnePlus AI + Gemini | 45.0 | Research required |
| 27 | OnePlus 13 | OnePlus AI | 45.0 | Research required |
| 28 | Honor Magic V4 | MagicOS AI | 45.0 | Research required |
| 29 | Honor Magic6 Pro | MagicOS AI | 45.0 | Research required |
| 30 | Xiaomi MIX Flip 2 | Xiaomi HyperAI | 45.0 | Research required |
| 31 | Pixel 9 Pro XL | Gemini Nano | 45.0 | Research required |
| 32 | Pixel 10a | Gemini Nano | 45.0 | Research required |
| 33 | Pixel 9 Pro | Gemini Nano | 45.0 | Research required |
| 34 | Galaxy S25 FE | Galaxy AI | 44.0 | Evidence conflict |
| 35 | Galaxy S24 FE | Galaxy AI | 44.0 | Research required |
| 36 | Galaxy Z Flip8 | Galaxy AI | 44.0 | Research required |
| 37 | Galaxy S24+ | Galaxy AI | 44.0 | Evidence conflict |
| 38 | iPhone 16 Pro | Apple Intelligence | 35.0 | Secondary source |
| 39 | iPhone 16 Plus | Apple Intelligence | 35.0 | Secondary source |
| 40 | iPhone 16 | Apple Intelligence | 35.0 | Secondary source |
| 41 | iPhone 15 | Apple Intelligence | 35.0 | Secondary source |
| 42 | iPhone 17e | Apple Intelligence | 35.0 | Research required |
| 43 | iPhone 17 | Apple Intelligence | 35.0 | Research required |
| 44 | iPhone 16 Pro Max | Apple Intelligence | 35.0 | Secondary source |
| 45 | iPhone 16e | Apple Intelligence | 35.0 | Secondary source |
| 46 | iPhone 15 Pro Max | Apple Intelligence | 35.0 | Secondary source |
| 47 | iPhone 15 Pro | Apple Intelligence | 35.0 | Secondary source |
| 48 | iPhone 15 Plus | Apple Intelligence | 35.0 | Secondary source |
| 49 | Galaxy Z Flip4 | Galaxy AI | 26.0 | Research required |
| 50 | Galaxy Z Fold5 | Galaxy AI | 26.0 | Research required |
| 51 | Galaxy Z Flip5 | Galaxy AI | 26.0 | Research required |
| 52 | Galaxy S23 | Galaxy AI | 26.0 | Research required |
| 53 | Galaxy Z Fold4 | Galaxy AI | 26.0 | Research required |
| 54 | Galaxy S23+ | Galaxy AI | 26.0 | Research required |
| 55 | Galaxy S23 Ultra | Galaxy AI | 26.0 | Research required |
| 56 | Galaxy S23 FE | Galaxy AI | 26.0 | Research required |
| 57 | Honor 400 Pro | MagicOS AI | 20.0 | Research required |
| 58 | Honor 400 | MagicOS AI | 20.0 | Research required |
| 59 | Pixel 10 Pro Fold | Gemini Nano | 18.0 | Research required |
| 60 | Pixel 10 | Gemini Nano | 18.0 | Research required |
| 61 | Pixel 10 Pro | Gemini Nano | 18.0 | Research required |
| 62 | Pixel 10 Pro XL | Gemini Nano | 18.0 | Research required |
| 63 | Pixel 7 Pro | Gemini Nano | 13.0 | Research required |
| 64 | Pixel 8 | Gemini Nano | 13.0 | Research required |
| 65 | Pixel 8a | Gemini Nano | 13.0 | Research required |
| 66 | Pixel 8 Pro | Gemini Nano | 13.0 | Research required |
| 67 | Pixel Fold | Gemini Nano | 13.0 | Research required |
| 68 | Pixel 7 | Gemini Nano | 13.0 | Research required |
| 69 | Pixel 11 | Gemini Nano | Not publicly specified | Not publicly specified |
| 70 | Pixel 11 Pro | Gemini Nano | Not publicly specified | Not publicly specified |
| 71 | Pixel 11 Pro XL | Gemini Nano | Not publicly specified | Not publicly specified |
| 72 | iPhone 18 Pro | Apple Intelligence | Not publicly specified | Research required |
| 73 | Pixel 11 Pro Fold | Gemini Nano | Not publicly specified | Not publicly specified |
| 74 | Galaxy S26 | Galaxy AI | Not publicly specified | Research required |
| 75 | Galaxy S26 FE | Galaxy AI | Not publicly specified | Research required |
| 76 | iPhone 18 Pro Max | Apple Intelligence | Not publicly specified | Research required |
| 77 | iPhone Duo | Apple Intelligence | Not publicly specified | Research required |
| 78 | Samsung Galaxy A27 5G | Galaxy AI | Not publicly specified | Research required |
| 79 | Galaxy S26 Ultra | Galaxy AI | Not publicly specified | Research required |
| 80 | Xiaomi Redmi Note 15 Pro 4G | Xiaomi HyperAI | Not publicly specified | Research required |
| 81 | Xiaomi Redmi Note 17 Pro Max | Xiaomi HyperAI | Not publicly specified | Research required |
| 82 | Samsung Galaxy A18 4G | Galaxy AI | Not publicly specified | Research required |
| 83 | Xiaomi Redmi Note 17 5G | Xiaomi HyperAI | Not publicly specified | Research required |
| 84 | Poco F9 Pro | Xiaomi HyperAI | Not publicly specified | Research required |
| 85 | Poco F9 Ultra | Xiaomi HyperAI | Not publicly specified | Research required |