Unverified50% confidenceOpinionExact time
综合实测结果，四款模型能力排名为：Sol > GPT-5.5 > Terra > Luna
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/24/2026
First Seen
Valid until: 10/22/2026
Sources
GPT-5.6三模型实测对比：Sol/Terra/Luna选哪个？
bilibili麦冬AI实验室7/11/2026
Related Claims
Unverified据Artificial Analysis数据，Qwen 3.8的270亿参数模型在Agentic指数上拿到51分，超过GPT-5.6 Terra、DeepSeek v4 Pro、GPT-5.6 Luna以及Gemini 3.7 Flash70% similarUnverified基准测试结果显示，Sol和Sol Ultra在大多数领域优于竞争对手Misos 5，Tara优于Fable 566% similarUnverifiedGPT-5.6 Sol在Humanity's Last Exam取得最高分，且价格远低于Claude Opus和Fable66% similarUnverified根据Artificial Analysis Arena的排名数据，xAI的Grok 4.6在综合能力上被认为与Sol 5.6处于同一水平66% 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%64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/611088API
curl https://kongchang.com/api/v1/knowledge/claims/611088MCP
get_claim(id=611088)