Unverified50% confidenceReleaseExact time

Zhipu's GLM 5.3 High Speed version has started gray-scale testing, with users measuring nearly 500 tokens per second and a first-token latency of under 5 seconds; it currently uses a random distribution mechanism and is not yet fully open.

1

Sources

50%

Confidence

Short-term (~14 days)

Relevance

10/13/2026

First Seen

Valid until: 10/27/2026

Sources

Alibaba Open-Sources QwenGyre: Long-Task Training Without Dropped Connections, Agents Learn While They Work

Related Entities

Zhipu AIcompany GLM 5.3 High Speedmodel

Related Claims

UnverifiedAccording to user measurements, GLM-5.3 High Speed generates at about 493 tokens per second, more than twice the peak of GLM-5.3 Flash.71% similar UnverifiedGLM-5.3在六项评测中的成绩为：终端编程基准3.0得28.3、深度软件工程基准得66.9、智能体中场考试得28.5、自动化基准得48.2、带工具的人类终极考试得62.5、GDP价值评估AA2版得17.6967% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/1014649

API

curl https://kongchang.com/api/v1/knowledge/claims/1014649

MCP

get_claim(id=1014649)