Unverified90% confidenceFactTime unknown
Claude Code 消费者用户（Free/Pro/Max）允许数据用于模型改进时，Anthropic 会在训练流水线中保留数据最长 5 年
1
Sources
90%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe knowledge cutoff of mainstream models often lags behind their release date by half a year to more than a year, because training data collection and cleaning takes several months.57% similarUnverified主流大模型的训练数据通常有6-18个月的滞后期57% similarUnverifiedNewly disclosed vulnerabilities typically require a 6–18 month data collection and cleaning cycle before entering training data.56% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/18171API
curl https://kongchang.com/api/v1/knowledge/claims/18171MCP
get_claim(id=18171)