Unverified50% confidenceFactExact time
Third-party platforms were less affected by the incorrect routing: on Amazon Bedrock, misrouted traffic peaked at 0.18% of all Sonnet 4 requests; on Google Vertex AI, fewer than 0.0004% of requests were affected between August 27 and September 16, 2025.
1
Sources
50%
Confidence
Long-term
Relevance
10/19/2026
First Seen
Sources
Related Entities
Related Claims
Unverified2012年Hinton团队的AlexNet以top-5错误率从26%降至15%的成绩，ResNet（2015，微软研究院）通过残差连接解决深层网络梯度消失问题，EfficientNet（2019，Google Brain）通过复合缩放系数优化网络64% similarUnverifiedThis developer's token usage by model is distributed as: Sonnet 82%, Opus 14%, Fable 3.6% (almost entirely concentrated within one day).61% similarUnverifiedThe margin of error for app download and revenue estimates derived by third-party data platforms (such as Sensor Tower, data.ai, and RevenueCat) is typically within ±20%.61% similarUnverifiedFor the same failing test fix, Haiku costs only about one-sixth of a cent—17 times cheaper than Sonnet—and completes in 6 seconds (compared to 10 seconds for Sonnet).59% similarUnverifiedThe improvement in agentic capabilities is accompanied by exponential growth in token consumption, and Sonnet 5's token consumption in maximum-effort mode is comparable to OPUS but with worse results.58% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1072485API
curl https://kongchang.com/api/v1/knowledge/claims/1072485MCP
get_claim(id=1072485)