DBS is matching different AI models to different tasks as it looks to contain costs while usage grows.
Chief Executive Officer Tan Su Shan told Bloomberg that heavier AI usage has been accompanied by lower costs per token as the bank improves how it deploys the technology.
Simpler requests can be handled by smaller language models, while DBS tests technology from several providers rather than relying heavily on one.
The bank also uses caching to avoid processing identical queries repeatedly.
Tan said employees are becoming better at optimising how the technology is used.
Despite wider adoption, DBS’ overall technology spending remains at around 10% of revenue and has been broadly stable, according to Tan.
The bank is also assessing which models and technology should be developed internally. Its open approach allows it to test different providers for specific uses.
DBS recently reported record second-quarter net profit of S$3.08 billion and raised its full-year outlook.
Featured image: Edited by Fintech News Singapor, based on image by starmultikharisma via Magnific
