Unverified50% confidenceBenchmarkExact time
GPT-6.1 Sol「工具出故障时不如实告知」比例从 4.9% 降到 2.1%，Astra 为 1.5%
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/30/2026
First Seen
Valid until: 12/29/2026
Sources
Related Entities
Related Claims
UnverifiedGPT-6 Sol在智能度几乎与上一代持平的前提下，价格下降了一半72% similarUnverified在ARC-AGI-3基准上，GPT-5.6 Sol得分8%，其他模型普遍不足2%，Anthropic因成本过高未运行Fable71% similarUnverified在96%的关卡里，GPT-6找到答案所需的尝试次数比普通人还少69% similarUnverified在Artificial Analysis智能指数综合测试中，GPT-5.6性能比Claude Opus 4.5略低约0.9分，但成本几乎减半67% similarUnverified在电脑操作类任务上Astra最强，Sol得分60.5%与Opus 5的60.3%接近，单任务成本约低八成66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/961320API
curl https://kongchang.com/api/v1/knowledge/claims/961320MCP
get_claim(id=961320)