Unverified75% confidenceFactTime unknown
GLM-4.6在某些任务中存在Token使用效率偏低的问题，用词量偏多
1
Sources
75%
Confidence
Medium-term (~90 days)
Relevance
6/1/2026
First Seen
Valid until: 8/30/2026
Sources
GLM-4.6深度实测：性能、价格与编程能力全面评估
bilibili薛定猫AI
Related Entities
Related Claims
UnverifiedGLM5.1相比GLM5在解决编程问题时所需的交互提示次数明显减少78% similarUnverifiedGLM5.1在复杂任务中高级功能的Token消耗速度远超预期74% similarUnverified在提示词中要求GLM5减少思考可以显著缩短推理过程并提升输出准确性73% similarUnverifiedGLM 5.1 High Speed声称在不降低模型能力的前提下实现400 token/s，突破点在于推理系统层面的工程优化而非模型压缩72% similarUnverifiedGLM5.1在长上下文场景下稳定性不足，多轮对话后容易出现前后矛盾的代码修改和逻辑混乱70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/24194API
curl https://kongchang.com/api/v1/knowledge/claims/24194MCP
get_claim(id=24194)