Unverified70% confidenceFactExact time
V4.1 Flash是原生多模态设计，能够直接看图理解内容
4
Sources
70%
Confidence
Medium-term (~90 days)
Relevance
9/11/2026
First Seen
Valid until: 12/10/2026
Sources
Related Entities
Related Claims
UnverifiedV4-Flash-Vision-Exp在纯文本Agent能力上与V4-Flash基本持平79% similarUnverified该模型基于V4 Flash架构，通过持续训练（continued training）叠加视觉模块获得图像理解能力79% similarUnverified在图片转网页测试中，V4.1-Flash基本还原了头像、统计信息和图形结构，UI还原精度相比V4.0-Flash有明显提升73% similarVerifiedDeepSeek将实验性多模态模型V4-Flash-Vision-EXP上线API平台，其文本能力与V4 Flash持平73% similarUnverified据B站UP主九天实测，DeepSeek V4 Flash搭配DSH V0.1.1的原生视觉能力可产出前端页面、HTML版PPT，并一句话生成接近商业级水准的宣传短片73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/908259API
curl https://kongchang.com/api/v1/knowledge/claims/908259MCP
get_claim(id=908259)