Unverified50% confidenceBenchmarkExact time
The analyzers executed by the policy engine hit the answer + evidence contract in all 110 out of 110 tests.
1
Sources
50%
Confidence
Long-term
Relevance
9/11/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedIn actual testing, one Deep Research call invoked 104 agents for targeted research, completing fact-checking based on 22 sources and 59 claims.68% similarUnverified每个模型配置以HIGH和XH（Extra High）两种推理强度进行测试，每种配置独立运行10次66% similarUnverifiedAgent可靠性验证方法论包括：每个任务重复运行10次以上、以程序化方式检查最终状态、对每次工具调用按schema校验、统计高负载下被截断的调用数量63% similarUnverified该评测共108项检查，分三大模块：后端API与全线65分、前端页面与真实交互25分、运行构建打包10分61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/898640API
curl https://kongchang.com/api/v1/knowledge/claims/898640MCP
get_claim(id=898640)