Unverified50% confidenceOpinionTime unknown
Gemini 3.1 Pro容易产生幻觉，有时输出质量不稳定
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
6/1/2026
First Seen
Valid until: 8/30/2026
Sources
Gemini 3.1 Pro + Claude Opus 4.6：AI编程最强组合实战指南
bilibiliAI-seeker
Related Entities
Related Claims
UnverifiedGemini 3在SWE-bench基准测试中智能体场景下的得分远超同类模型78% similarPartially VerifiedGemini 3.0 Pro的前端还原度被评为全场第一，但后端错误率较高77% similarUnverifiedGemini 3.2 Pro的UI生成能力表现薄弱，相比SVG方面的提升并不明显77% similarUnverifiedGemini 3.1 Pro发布后，不少开发者对其表现感到失望，认为单次实力测试大幅下滑76% similarVerifiedGemini 3.1 Pro新增了中档思考级别选项，而Gemini 3只有低和高两档76% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/22235API
curl https://kongchang.com/api/v1/knowledge/claims/22235MCP
get_claim(id=22235)