|
Work item:
|
Q.MP-LLM-FN
|
|
Subject/title:
|
Computing power resource measurement model and parameters for large language model inference in future networks
|
|
Status:
|
Under study
|
|
Approval process:
|
AAP
|
|
Type of work item:
|
Recommendation
|
|
Version:
|
New
|
|
Equivalent number:
|
-
|
|
Timing:
|
Q4-2027 (Medium priority)
|
|
Liaison:
|
ITU-T SG13, SG21
|
|
Supporting members:
|
China Unicom, China Telecom, China Mobile
|
|
Summary:
|
This draft Recommendation specifies a computing power resource measurement model and key parameters for large language model (LLM) inference in future networks. It introduces a three-layer hierarchical measurement model consisting of the LLM service layer, the computing power capability layer, and the computing power resource layer, and identifies the essential parameters at each layer. The objective is to provide a basis for resource estimation, allocation, and optimization of LLM inference services in future networks.
|
|
Comment:
|
-
|
|
Reference(s):
|
|
|
Historic references:
|
|
Contact(s):
|
|
| ITU-T A.5 justification(s): |
|
|
|
|
First registration in the WP:
2026-03-19 15:18:45
|
|
Last update:
2026-04-02 15:07:59
|
