Unverified60% confidenceBenchmarkExact time
在Terminal Bench 4.0最高精度测试中，Fable 5.1准确率为55.8%，成本为19.50美元；Astra准确率为56.7%，成本为10.35美元
2
Sources
60%
Confidence
Medium-term (~90 days)
Relevance
9/9/2026
First Seen
Valid until: 12/8/2026
Sources
Related Entities
Related Claims
Unverified在 Frontier Code 准确率与成本基准测试中，低档位花费约 5 美元可达约 11% 得分，而 Opus 4.8 Max 花费约 11 美元才拿到相同得分73% similarUnverifiedAS5048 数据手册中标注的实际精度（考虑积分非线性误差 INL）通常在 ±0.5 度左右69% similarUnverifiedGPT-6 Astra在256K-512K token区间的八针基准测试中准确率达到100%66% similarUnverified微调后的Qwen3.5-4B模型在BU Bench V1基准上准确率从15%提升至53%65% similarUnverifiedGPT-5.6 Extreme High档位约需63美元获得53.6%得分，而Claude Sonnet自适应模式耗费约2000美元得分仅40.5%65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/881470API
curl https://kongchang.com/api/v1/knowledge/claims/881470MCP
get_claim(id=881470)