Report
Revolut reportedly makes self-hosted GLM its default AI coding model
Revolut says its GLM bill is roughly a tenth of what it paid OpenAI and Anthropic.
TLDR
Revolut reportedly made GLM the coding default for about 2,000 engineers, using 56 Nvidia B200 GPUs while keeping Claude and GPT for select tasks. Revolut puts its blended cost at 30 cents per million tokens and says unlimited tokens replaced spending caps. The company says coding time fell 60% and engineering lead time fell 70%.
Combined views
20.7K
3 Sources, first seen ago
