Unverified50% confidenceBenchmarkExact time
In Harvey LAB-AA legal analysis, Muse Wonder achieved 13.1%, more than three times Claude Opus 5.5's 4.2%; in Vals Finance Agent V2 it ranked first with 59.1%.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/13/2026
First Seen
Valid until: 10/27/2026
Sources
Related Entities
Related Claims
UnverifiedArtificial Analysis gives Muse Wonder an intelligence index of 52, ranking 9th out of 219 models; Claude Opus 5.5 is 58, Astra is 53, and Wonder Full is 56.61% similarUnverifiedThe GLM-Legal model launched by Z.ai achieved a 19.2% all-pass rate on the Harvey LAB legal Agent benchmark, surpassing Claude Fable 5 and Kimi K3.58% similarUnverified在编程智能体测试中，Sol 输出的 Token 少了一半以上，成本约为 Faber-5 的 1/357% similarUnverified综合智能指数由智能体工作、编程、科学推理、通用能力等 9 项独立评估构成，Sol 与 Faber-5 几乎持平仅差 1 分，但任务完成时间减少约 61%，成本约为原来的一半56% similarUnverifiedSol scored 68.8% in the coding test, close to the previous-generation Fable 5's 69.9%, but with a per-task cost approximately 80% lower.56% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1020384API
curl https://kongchang.com/api/v1/knowledge/claims/1020384MCP
get_claim(id=1020384)