AVSSLP: A Six-Dimensional Evaluation Framework for Artificial Intelligence Systems

This paper introduces AVSSLP, a six-dimensional framework for evaluating the performance, efficiency, and resilience of artificial intelligence systems. The framework consists of six dimensions: Actions per Second (APS) Value of Actions Performed Sustainability Stress Test Latency Power Usage AVSSLP expands conventional performance evaluation beyond throughput and response time by incorporating action value, energy efficiency, and system behavior under demanding operating conditions. The framework proposes operational definitions, measurement approaches, experimental considerations, and reproducibility guidelines for each dimension. It also examines how these measurements can reveal trade-offs that conventional performance benchmarks may overlook. A central principle of AVSSLP is that increased computational activity does not necessarily represent increased useful performance. Systems must be evaluated according to the value of their completed actions, the resources required to produce them, and their ability to maintain performance under stress. AVSSLP is proposed as a general-purpose evaluation methodology applicable to AI models, agentic systems, inference architectures, and adaptive resource-management systems. This work establishes the theoretical and methodological foundation for future AVSSLP-based experiments.

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-14
DOI
https://doi.org/10.5281/zenodo.24213216
Primary Topic
Software System Performance and Reliability
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

AVSSLP: A Six-Dimensional Evaluation Framework for Artificial Intelligence Systems

Devin "River" Bigelow
Zenodo (CERN European Organization for Nuclear Research)
Software System Performance and Reliability
preprint

AVSSLP: A Six-Dimensional Evaluation Framework for Artificial Intelligence Systems

Devin "River" Bigelow
preprint en

Abstract

This paper introduces AVSSLP, a six-dimensional framework for evaluating the performance, efficiency, and resilience of artificial intelligence systems. The framework consists of six dimensions: Actions per Second (APS) Value of Actions Performed Sustainability Stress Test Latency Power Usage AVSSLP expands conventional performance evaluation beyond throughput and response time by incorporating action value, energy efficiency, and system behavior under demanding operating conditions. The framework proposes operational definitions, measurement approaches, experimental considerations, and reproducibility guidelines for each dimension. It also examines how these measurements can reveal trade-offs that conventional performance benchmarks may overlook. A central principle of AVSSLP is that increased computational activity does not necessarily represent increased useful performance. Systems must be evaluated according to the value of their completed actions, the resources required to produce them, and their ability to maintain performance under stress. AVSSLP is proposed as a general-purpose evaluation methodology applicable to AI models, agentic systems, inference architectures, and adaptive resource-management systems. This work establishes the theoretical and methodological foundation for future AVSSLP-based experiments.

Zenodo (CERN European Organization for Nuclear Research)
Software System Performance and Reliability
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon