Verified65% confidenceBenchmarkExact time
苹果M4芯片的Neural Engine算力达38 TOPS，高通骁龙8 Elite的Hexagon NPU达45 TOPS，英特尔Lunar Lake集成的NPU算力达48 TOPS
3
Sources
65%
Confidence
Medium-term (~90 days)
Relevance
7/7/2026
First Seen
Valid until: 10/5/2026
Sources
捍卫本地AI运行权：隐私、自由与数字主权之战
hackernewshackernews7/2/2026
Related Claims
UnverifiedApple M4 Pro的Neural Engine达38TOPS，高通骁龙X Elite集成45TOPS NPU，英特尔Core Ultra也已集成NPU支持本地AI推理83% similarUnverified苹果M系列芯片中的Neural Engine本质上属于NPU76% similarVerifiedApple Silicon的Neural Engine已能在M系列芯片上流畅运行数十亿参数模型，高通骁龙X Elite等端侧芯片专为AI推理优化75% similarUnverifiedApple Silicon（M系列）芯片对AI修图有原生优化，Topaz、Pixelmator Pro等针对Neural Engine优化的软件在M2/M3上性能接近中端独显PC，功耗却极低。Mac用户选购时应确认软件是否为Apple Silicon原生适配而非Rosetta转译运行74% similarUnverified苹果的Neural Engine（ANE）自A11 Bionic（2017年）首次引入，A17 Pro的ANE算力达35 TOPS，M4芯片的ANE达38 TOPS74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/185468API
curl https://kongchang.com/api/v1/knowledge/claims/185468MCP
get_claim(id=185468)