Unverified50% confidenceFactExact time
GLM 5.3 Flash 同时提供视觉（vision）与文本（text）两种模态的训练支持
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/4/2026
First Seen
Valid until: 1/2/2027
Sources
Related Entities
Related Claims
Verified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75% similarUnverifiedGLM-5.3-Flash采用MIT许可证开源，支持图像、Token等多模态输入72% similarUnverified谷歌的PaLI、OpenAI的GPT-4V、Meta的LLaMA Vision均沿用了类似的图文对齐预训练策略68% similarVerifiedV4.1 Flash是原生多模态设计，能够直接看图理解内容66% similarUnverifiedFlash更擅长前端任务和Pro本身不支持的实图能力65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/973253API
curl https://kongchang.com/api/v1/knowledge/claims/973253MCP
get_claim(id=973253)