Revolut says it cut AI coding inference costs by ~90% in less than a year by making self-hosted GLM models its default over commercial tools like Claude Code (Tobias Mann/The Register)
The company managed to slash its artificial intelligence programming inference expenses by around 90% within a single year. To realize these efficiencies, the firm transitioned its default development setup to self-hosted GLM models. This strategic move replaced reliance on external paid solutions, including commercial tools such as Claude Code. The transition demonstrates how businesses can effectively curb escalating development overhead through tailored, on-premises machine learning implementations.
Key points
Moving from proprietary cloud services to self-hosted models can dramatically curb softwa…: Moving from proprietary cloud services to self-hosted models can dramatically curb software development costs.
Managing artificial intelligence systems internally allows enterprises to scale developer…: Managing artificial intelligence systems internally allows enterprises to scale developer tools without incurring prohibitive vendor fees.
Organizations are adopting private open-architecture setups to gain control over their lo…: Organizations are adopting private open-architecture setups to gain control over their long-term infrastructure budgets.
Sources checked by our analysis pipeline: 1
New York Times: |
Jeff John Roberts / Fortune: |
Financial Times: |
Michael Acton / Financial Times: |
Apple: |
Dave Lee / Bloomberg: |
Sarah Perez / TechCrunch: |
Apple Developer: |
Bloomberg: |
Tom Wilson / Financial Times: |
Lance Whitney / ZDNET: |
Gareth Vipers / Wall Street Journal: |
Natasha Mascarenhas / Bloomberg: |
Alex Kantrowitz / Big Technology: |
Financial Times: |
Emma Roth / The Verge: |
James Rundle / Wall Street Journal: |
Jason Schreier / Bloomberg: |
Steven Lee Myers / New York Times: |
David Pierce / The Verge: |
Samantha Subin / CNBC: |
Katie Tarasov / CNBC: |
Colin Demarest / Axios: |
New York Times: |
Tobias Mann / The Register: |
Gerrit De Vynck / Washington Post: |
OpenAI: |
This is a Techmeme archive page. It shows how the site appeared at 4:50 PM ET, November 3, 2026.
The most current version of the site as always is available at our home page. To view an earlier snapshot click here and then modify the date…
Emerging technologies are reversing centuries of rigid office culture by driving a modern revival of decentralized, home-based labor. Advanced digital platforms and automation are cutting business overhead, allowing independent operators to manage complex ventures without conventional corporate infrastructure. This shift enables workers across generations to trade traditional employment for greater autonomy and self-directed schedules.
The company is grappling with severe liquidity issues after raising just $11 billion of an initial $30 billion target at a $1.25 trillion valuation. Backers have refused to commit the additional $9 billion needed to keep operations funded for the coming months. A potential insolvency risks setting off broader industry fallout, threatening to idle data centers and eliminate demand for graphics processing units.
Author Tianhua Chen has introduced a derivation-focused reference designed to explain the essential mathematical structures powering modern generative models. Rather than cataloging specific implementation nuances, the work helps mathematically inclined students, practitioners, and researchers understand how major framework families interconnect. This resource aims to provide clear conceptual access while maintaining rigorous formal derivations across foundational artificial intelligence techniques.