Unverified50% confidenceFactExact time
After deleting more than 80% of the system prompt, running the official benchmarks showed no decline in scores.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
8/29/2026
First Seen
Valid until: 11/27/2026
Sources
Related Entities
Related Claims
Unverified当所有模型在某评测上都能拿90分时该评测就应退场，即基准饱和现象70% similarUnverified研究表明，超过80%的审稿人在rebuttal后不会实质性更改评分67% similarUnverifiedAnthropic deleted more than 80% of Claude Code's system prompt.65% similarUnverified如果编辑工具的成功率只有90%，在需要连续执行10次编辑的任务中，整体成功率将下降到不足35%63% similarUnverifiedIn a well-calibrated model, an output score of 0.8 should statistically correspond to an actual accuracy of 80%.62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/820860API
curl https://kongchang.com/api/v1/knowledge/claims/820860MCP
get_claim(id=820860)