Unverified50% confidenceBenchmarkExact time
Haiku 5.5在Computer Use任务上比GPT-6 Luna高出约40%
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/9/2026
First Seen
Valid until: 1/7/2027
Sources
Related Entities
Related Claims
UnverifiedHaiku 5.5 在计算机使用（computer use）任务上的得分从 16% 跃升至 72%77% similarUnverified在max等高推理强度设置下，Haiku 5.5的token消耗显著增加，单任务成本可能比GPT-6 Luna更不划算73% similarUnverifiedHaiku 5.5在不同推理强度之间的能力跃升明显，从high到max/X-high有相当大的提升幅度，不像Opus 5.5或Sonnet70% similarUnverifiedHaiku 5.5的具体性能基准、定价细节与上下文窗口等关键参数尚未在公开信息中披露70% similarUnverified在Databricks的OfficeQA Pro V1评测中，Claude Haiku 5.5相比上一代Haiku 4.5实现了约15%的质量提升，同时成本仅为后者的一小部分70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/990911API
curl https://kongchang.com/api/v1/knowledge/claims/990911MCP
get_claim(id=990911)