Unverified60% confidenceEventTime unknown
Gemini 3.2 Pro的首批泄露测试结果已经曝光，SVG生成能力有所提升但整体表现并不令人惊艳
1
Sources
60%
Confidence
Medium-term (~90 days)
Relevance
6/3/2026
First Seen
Valid until: 9/1/2026
Sources
Gemini 3.2 Pro泄露测试令人失望，GPT-5.6已在内部测试
bilibiliAI-seeker
Related Entities
Related Claims
UnverifiedGemini 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84% similarUnverifiedGemini 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% similarUnverifiedGemini 3.1 Pro在资产生成（SVG、精灵图、动画）方面具有明显优势76% 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76% similarVerifiedGemini 3.5 Flash在几乎所有基准测试中超越了Gemini 3.1 Pro75% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/38416API
curl https://kongchang.com/api/v1/knowledge/claims/38416MCP
get_claim(id=38416)