Unverified50% confidenceFactExact time
GLM-5.2 团队大幅扩展了面向编码智能体场景的 1M 上下文训练，覆盖大规模实现、自动化研究、性能优化和复杂调试等任务类型
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
8/26/2026
First Seen
Valid until: 11/24/2026
Sources
Related Entities
Related Claims
Unverified智谱GLM-5.3基于743B（7430亿参数）基座进行后训练，主打编程与智能体两大核心场景77% similarVerifiedThe GLM series originally adopted an autoregressive blank-filling pre-training framework that accommodates both understanding and generation capabilities.76% similarUnverifiedGLM-5.2 引入了多档思考努力等级（thinking effort levels），允许开发者在性能与延迟之间权衡71% similarUnverifiedMegatron-LM引入序列并行策略，针对LayerNorm、Dropout等算子将序列维度切分至多GPU处理，提升超长上下文训练场景下的显存利用率70% similarUnverifiedZhipu AI is a technology commercialization spinoff of Tsinghua University, and its GLM architecture adopts an autoregressive blank-filling pre-training objective.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/805119API
curl https://kongchang.com/api/v1/knowledge/claims/805119MCP
get_claim(id=805119)