Verified65% confidenceBenchmarkExact time
The M4 chip's Neural Engine delivers a peak compute performance of 38 TOPS (38 trillion operations per second)
3
Sources
65%
Confidence
Long-term
Relevance
8/29/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedApple Silicon Neural Engine的峰值算力从M1的11 TOPS跃升至M4的38 TOPS，性能提升超过3倍78% similarUnverified苹果A18 Pro芯片的Neural Engine峰值算力达到约35 TOPS，足以本地运行30亿至70亿参数量级的压缩语言模型76% similarUnverifiedArtificial Analysis给Muse Glimmer 30B打出35分的智能指数，相比Llama 4 Maverick是巨大飞跃65% similarUnverifiedQwen3 27B achieved a 100% pass rate in the 256K long-context memory retrieval test (needle in a haystack), with all 15 runs being accurate.64% similarUnverifiedThe model parameters required to run agents on an agentic host need to reach at least 30 billion (30B–35B) to be sufficiently intelligent.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/817905API
curl https://kongchang.com/api/v1/knowledge/claims/817905MCP
get_claim(id=817905)