Unverified50% confidenceFactExact time
GLM-5系列从头到尾在华为昇腾910B芯片上训练，大约使用了10万块，配套华为自研的MindSpore框架，全程零英伟达依赖
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/2/2026
First Seen
Valid until: 9/30/2026
Sources
GLM-5.2深度解析：百万上下文、MIT开源与国产算力全栈突破
bilibiliAI技术投降派6/14/2026
Related Claims
Unverified研究团队成功使用华为Ascend 910C芯片完成DeepSeek v4 Pro模型的后训练72% similarUnverified智谱模型据称在华为昇腾910D芯片上完成训练，910D目标对标英伟达H10071% similarUnverifiedGLM-5和GLM-5.1已完成华为昇腾、摩尔线程、寒武纪、昆仑芯、沐曦、燧原、海光7大国产芯片的推理适配70% similarUnverifiedZhipu AI stated that GLM-5.2 was trained on full-stack domestic chips.69% similarUnverified华为构建了昇腾芯片、MindSpore框架、盘古模型、鸿蒙系统、麒麟芯片五件套全栈自研生态66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/46082API
curl https://kongchang.com/api/v1/knowledge/claims/46082MCP
get_claim(id=46082)