GoKawiilRevolut switches ~2,000 engineers to self-hosted GLM for coding tasks
Revolut has migrated roughly 2,000 engineers to a self-hosted GLM model for coding, running it on 56 Nvidia B200 GPUs, according to David Abitbol, who leads AI operations at the company. Claude Code and Codex, previously the default tools from Anthropic and OpenAI, have been moved to an exception list, with Abitbol citing a blended cost of about 30 cents per million tokens, roughly a tenth of prior spend.
GoKawiil's interpretation of the reporting above, not reported fact.
Abitbol reports a 70% improvement in engineering lead time, which he attributes to engineers no longer rationing AI usage once costs fell, instead applying it broadly rather than saving it for difficult tasks. This case suggests that cheaper self-hosted open-weight models could pressure frontier AI labs on pricing as they approach IPOs, though it remains a single company's experience rather than proof of a broader trend.
- Revolut shifted ~2,000 engineers from Claude Code/Codex to self-hosted GLM on 56 Nvidia B200 GPUs
- Blended cost dropped to about 30 cents per million tokens, roughly a tenth of previous spend
- Engineering lead time improved 70%, which Abitbol links to removing token-usage rationing
Source: twitter.com, 2026-11-03
Published there as: “Revolut moves to self-hosted GLM”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.