When time-series data is rendered as plots, vision-language models use 3.6 to 10.4x fewer input tokens and 1.8 to 2.5x less measured inference energy than text-only LLMs, with equal or better anomaly detection accuracy.
Artificial intelligence and energy overcon- sumption: Data center electricity demand, cooling burdens, and regional sustainability constraints,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.AI 1years
2026 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
A Picture is Worth a Thousand Tokens: How Vision Language Models Cut AI Energy Costs While Improving Accuracy
When time-series data is rendered as plots, vision-language models use 3.6 to 10.4x fewer input tokens and 1.8 to 2.5x less measured inference energy than text-only LLMs, with equal or better anomaly detection accuracy.