Unverified50% confidenceBenchmarkExact time
在Agents' Last Exam评测中，GPT-5.6 Sol取得53.6分，比采用自适应推理的Claude Fable 5高出13.1分
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/13/2026
First Seen
Valid until: 10/11/2026
Sources
GPT-5.6三款新模型发布：Luna、Terra、Sol主打智能体长任务
rss7/9/2026
Related Claims
Unverified在长周期智能体工作流测试（Engines Last Exam）中，GPT-5.6 Sol 得分 53.6，比 Faber-5 高出 13.1 分84% similarUnverifiedGPT-5.6在Agent's Last Exam测试中以53.6分创下历史新高，领先Claude Opus 4.5达13.1分82% similarUnverified在Artificial Analysis的Omniscience幻觉指标上，GPT-5.6 Sol Max高达89%，高于Claude Fable的55%和GLM 5.2的28%81% similarUnverified在SWE-Bench Pro编程评测上，Claude Fable 5以80%的成绩领先GPT-5.6 Sol的64.6%81% similarUnverified在病毒学多选基准上，Mythos 5得分56%，GPT-5.6 Sol得分55.5%，几乎持平且均远超专家基线78% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/497874API
curl https://kongchang.com/api/v1/knowledge/claims/497874MCP
get_claim(id=497874)