Unverified50% confidenceFactExact time
V4-Flash-Vision-Exp能够读取截图、解析图表，并结合工具调用完成多模态Agent任务
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/3/2026
First Seen
Valid until: 12/2/2026
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78% similarUnverified视觉AI涵盖目标检测、图像分割、OCR、视频分析等核心技术，典型技术栈包括PyTorch/TensorFlow、OpenCV以及ONNX/TensorRT等推理优化工具73% similarUnverified采用自研影像芯片V3配合蔡司光学系统，视频拍摄支持4K电影感模式，定位是拍照与视频双强的影像旗舰73% similarUnverifiedDeepSeek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% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/849884API
curl https://kongchang.com/api/v1/knowledge/claims/849884MCP
get_claim(id=849884)